Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
AI x Crypto crossovers: What role will DID and ZKP play?
Recorded: June 19, 2025 Duration: 1:02:12Space Recording
Short Summary
In a dynamic discussion, key players in the crypto space explored emerging trends in decentralized identity, significant funding rounds, and innovative partnerships aimed at enhancing user privacy and security in the evolving Web3 landscape.
Full Transcription
Thank you. Thank you. It looks like it's not going to work.
I'm trying to speak for my personal account, and it's failing again.
How are you doing today, Jeff?
I'm pretty good, apart from we are so hot in the UK at the moment and we're not used to this weather
you're hot in terms of like you guys are keeping it spicy or you're hot in terms of like
you know bikini weather a bit of both a bit of both um but we're up in your temperature, Humpty. We're up at 86 degrees.
Oh my God, that's so pleasant.
How cute.
We're melting.
We're either in the beer garden or melting at home, one or the other.
Oh man, tell me your weather is beautiful or at least cooler most of the time without telling me.
I think that was a confession you just made.
Oh yeah, this is not... Follow me. I think that was a confession you just made. Oh, yeah. I think compared to our average temperature, we're like 15, 18 degrees higher than usual.
So where are you at normally, like 72 Fahrenheit?
Yeah.
For us Americans who don't know Celsius.
So I think that's right. Let me just put it into my calculator so I can understand what you're saying to me.
Oh, no, less than that. Less than that.
Stop it.
So right now we'd probably be about 66.
That would be your weather? Wait, OK, what's the hottest?
Now we're pretty much at our hottest.
So the hottest you would normally get during the peak of summer is around 86 Fahrenheit?
Yeah, that's right.
That's adorable.
It's so hot.
I'd say, okay.
I choose to live in California, and I choose to live in Southern California to be more exact,
And I choose to live in Southern California to be more exact, where it's a desert.
where it's a desert.
And we've just created a lot of huge cities in the desert of California.
And our peak most summers is definitely cracking 100, usually 108, sometimes 116.
That's insane i'd need to i'd need to live in a swimming pool well air conditioning is a requirement here you know if you don't have it wow like that's the
thing right so we're not we're not built for it we've got very well insulated we're in houses that are built to keep the heat in
yeah i live in a very old house and insulation is is not great meaning the cooler months it gets
cold the hotter months it gets hot so I do need to work on the insulation.
That's number one.
Two, though, I did install an air conditioner system about five years ago.
And that just changed our entire lives.
Yeah.
I keep thinking about it.
Because summer is getting hot in the UK at the moment.
You know, we get some quite hot summer, so it might be worth us doing something.
Yeah.
No, we won't talk about global warming because we'll just scare half the crowd away.
That's true.
That's true.
Because that's not real.
All right.
Cool.
Well, let's kick it off.
I know we were expecting to have a few other folks joining us. I just want to acknowledge some of the people that we were supposed to have up here and more so as a hopefully callback when we do have them up and maybe ask them to share their thoughts on some of the things we talked about today but one of them is sam coming back from last week actually uh from vera actually
that was a really great conversation if you missed it i definitely recommend going playing it back
um i think what vera is building you know it's like the new brave browser i would say
i know that that's probably not the best description of it, but most of us are familiar
with Brave browser, I would say. This really takes it above and beyond where that ended, where
you explore not just the traditional web, but the new web, right? Web3 in that browser. So you have decentralized applications
running in browser, you have a built in wallet, loyalty points
and systems in which you can accrue rewards, like a lot of
the stuff that like, I've been very excited about, when it comes
to kind of web three tech, and even worked on to some degree,
like with orange protocol, and then eventually with mosaic,
which was supposed to be like this loyalty platform.
I just ran out of money.
Hard to bootstrap stuff when you're a solo founder, solopreneur.
Yeah, it's just this is all built into Vira.
So I'm very excited about what they're building and excited to have Sam back in the future.
Then also we were supposed to have someone that you met, Jeff, from Terminal 3. Tell
us a little bit about that. Yeah, so they're an interesting project, right? So I still spend a
little bit of time being old school over on LinkedIn and doing some things over there.
Bro, LinkedIn is new school. It is not old school. No, it's going through a resurgence. They're
calling it the new Facebook app because of how well it performs socially.
And they haven't lost the plot like Meta did with ads and all the other crap they're doing to really make their platforms unusable.
So, no, you're doing great.
You're actually a Gen Z-er now.
You were laughing at my hair the other day because I just got my curly mullet in.
No, wait, okay.
I was laughing, but then the last meeting, sorry, I keep interrupting you.
You look very handsome.
You cleaned it up.
I don't know if you put gel or what, but you look great.
I just got up a little bit earlier.
That's all.
And I think eventually these things just come back into fashion, right?
So I've been in LinkedIn long enough that it's come back into fashion.
That's the key here. And same with the mullet you know it's just you have to
wait long enough i even got a little mustache at the moment you do okay look by the way for anybody
who hasn't met me i'm bald and can't grow facial hair so i 100 will be jealous of anybody who has
hair beautiful locks like jeff, and can grow facial hair.
Because, man, I grow this thing out for over a month,
and it's just like these little strangling pieces of hair stuck to my face.
You've just got to persevere, Humpkin.
There's just been very long, strangling pieces of hair on my face.
So they're a really interesting project, right?
So they're based in Hong Kong.
They're in identity.
They're working in identity and the AI space as well.
So that makes them an interesting prospect as well.
They just got, I think, 8 million funding in April for their seed round
or some funding round.
I'm not sure which version it was.
So, yeah, just a quick chat so far.
I'm going to hook up with them in cans as well at ECC
and have a chat with them there.
So interesting.
It's always exciting to see people coming into the space
and building things out in the identity space.
Love to hear that.
So I look forward to seeing them.
Maybe the next space we'll coordinate with them
with a little bit more time.
Excuse me, so that they can't come here.
Also, I met someone maybe about a week or so ago through another group chat that I'm in.
What is it called?
It's a Slack server kernel.
They are a community that and really people who are building in the public good space they help with leadership
skills and so on and i'm in this channel called jam identity and someone posted there about two
weeks ago that they were looking for institutional players in the decentralized
identity space. I reached out, I talked to them, I actually was on a call with them about an hour
or two ago, and we dove into institutional DID plays and really kind of the work that they're doing to standardize and implement some of these self-sovereign identity systems
and zero-knowledge proof systems, which we've been talking about at length here now for the last couple of weeks.
And we'll talk a little bit more about it today.
And, you know, and so some of these companies they were mentioning were like Salesforce and Microsoft.
Obviously, I got really excited about it.
And so I wanted her to join us, but I might have remembered to invite her a little too late.
So 30 minutes ago, I was like, by the way, we're hosting a space today, but she might be asleep.
She's closer to your neck of the woods.
She also mentioned there's going to be a DID or some self-sovereign identity event that's happening in Germany.
And the Decentralized Identity Foundation is co-organizing it.
Might be a good stop if you're already out in Paris, maybe on your way back home.
To check that out, I'll send you more info, Jeff, and see if we can get you out there.
Yeah, drop it over.
I mean, Germany is a really easy journey for me.
It's no problem at all.
Awesome.
Well, and then we do want to recognize the people that did show up.
You know, hell, we're over here, you know, complaining, not complaining, but like, you know know celebrating the people who are not here but
definitely was you're back and actually quite relevant you just recently tweeted something
and that jeff brought to my attention and i think uh it's really really great to hear like what
you're working on at least what you're envisioning would you mind just giving us a brief introduction
again to yourself for those who are newly tuning
But also, what is it that you tweeted or even put it on the bulletin board here so we could
all see it?
Thank you, Hamte.
Yeah, Jeff and me, we met each other, I think, a few months ago on Twitter randomly.
By the way, let me know about Germany, which city this event is having because I'm based
in Germany.
Oh, yeah. You know what I'll do is I'll send it to the group chat because I don't have it in front of me right now.
For sure. Yeah. So I started my journey in Web3 in 2016 as a gamer moving out from Maldives to coming to Europe and getting stuck with the money, like $50, $100, and I have to stage it up to $1,000 for the Deutsche Bank so I can send it to Maldives and India and Philippines to my streamers.
That's how in 2016 I came into Web3.
And fortunately enough, I was about to connect some people in Discord
and went into the whole jargon or the trenches of the Web3
and decided to leave my text and all this paid job
in germany and went full-time actually into web 3 in 2020 uh that was in darwinia ecosystem in the
evaluation land when the nft then lands and sells and all these things come up and in 2021 started
working for cd5 as the very first business development manager launching leading as a
business development manager to launch 21 projects that's including Blocktopia, Sidious, Arta and so on and so forth.
But soon enough as a gamer, what I understood is we need, I had been
exploring the DID for a very long time because what you need is the thing we
are talking about, that privacy and having the DID and all these things.
If you are a streamer, you don't definitely want to
know how many Robux has been paid by the Roblox for you to go to fill on their stream. That's one
thing. The other thing is you need to have a serious identity or serious solution that will
onboard you and all these things. Yes, that was running behind my back of my brain. And what
happened was I got hacked. And the best solution normally comes if you
find the solution for yourself. In 2023, when I was trying to build something that was
related to gaming and all these things, I got hacked. And then I started looking for
solutions. Already wallets exist, why nobody is building a wallet that allows people to
do two signatures and all these things. Okay, if that was done to devices, people from the emerging countries might not have
two devices. But that was not the best solution. So since
2023, 2024, me and another friend, we had been exploring
existing solutions, not to revamp the whole thing, but
existing solutions, what we can do. And then actually, I met
Geoff, had the Twitter space, I think I was three and a half, a few months ago, I think, last year somewhere.
And Ontology DID, we talk about at the time you guys were launching Orange, I think.
And then I was having Twitter space with you guys, Jeff, I think with Gamion Forge, that was.
And then this is where we started exploring the whole DID solutions.
What it could do for people to avoid connecting wallets in the ecosystem.
Like a new project comes in asking people to connect the wallet. And in that case, what the DID can play as a role to avoid that.
So the best solution we could find is having a burner wallet and having mapped a burner wallet to mint mint to the ID and map that the ID using IPF map that the ID to the other
wallets that you manually submit to your profile which you are on onboarding into
your profile attached into your profile which you will never do assign on my
website as well about the website we are creating the the the mapping and all
these things and then having an SBT connected to this DID, and then
using ERC 721, and then using IPFS, we basically have this SBT giving the user behavior to
update the merit system.
The reason I'm going to do this is because when I found a solution, me and my friend
we found a solution, we wanted to give it additional utility because we don't want to launch a token. We don't want to raise
the funds, even currently I work for TenSet, my profile is down there, TenSet,
one of the venture capital, Alphablux Ventures and all these things. If I
go and launch the token and all these things raise the funds, it will be again
the same thing all over again. So then we were looking for what we could do that
actually can give additional value to the users.
And then what we are right now doing is InfoFi.
And a person coming from marketing with hospitality background,
the current marketing or InfoFi VC is not really InfoFi.
It's actually compiling existing data of social media,
compiling it together and creating a buff or demand on top of that data to create more junk,
which is noise, not signal.
If you're looking for a signal like Bloomberg Terminal,
you need to have the signal or important information
or decision-making information that you can make according to educated guesses
that happened in previous history to predict the future.
That's basically the signal kind of thing.
And then what you need to do is the capital movement.
And how do we do this? So we said, okay, if Humpty is coming to my profile,
make my protocol come in and creating a burner wallet, connecting his other worm, we call them
worm wallets, because that's not soft wallets, not cold wallets like you have in the hardware
and all these things, but worm wallets that you never do assign on an index or even in our protocol now we know how much money or how much asset you have what kind of assets
and liquidity and all these things now what we do is basically we know the capital movement
so what we wanted to do is what's the extracted value that happened in web3 and then what we see
is fundraising for web3 founders when they go to fundraise, accelerate the project, a launchpad and all these things. If you are a launchpad you ask people to hold
the token and then during the time a project raises funds $200,000, $300,000
people are holding the token market makers keep on market making on one side
the project raises $200,000 the launch the whole token who hold the token makes
around $400,000 in market making.
We don't have to go to the juggernaut.
And then if you are in community venture capital, you are paying 10 to 15%.
So we thought, okay, that would be the first solution.
That would be the 1% of the utility that I can add to the value, which already I have
started because my network of the opportunities that I can bring to the community is huge
as working in venture capital and those things.
So right now what I started is bringing those people, bringing exclusive deals, bringing creator deals to these people without charging any penny from them.
So project, if it's a creator deal, project gives directly to the people.
If it's an investment opportunity, project also collaborates directly with the project users.
So user already gets some kind of value by creating this DID. That's one
thing and in the future what I want to do in this with this DID is now I know
the user capital then the other thing I want to use is the API of Twitter which
is what Kaito is doing that now pulled it up and give a profile to the user that
shows if he was given an airdrop what did he do and determining
if he sold it if he is if he sold it but still if he is using the decks for
example ontology is a deck and there was a deck's airdrop given and the person
sold the airdrop but still he was spending two thousand dollars monthly on
Uniswap but now he's spending one thousand dollar on ontology decks and
the other one thousand dollar on Uniswap so even if he's a person who sold for profit,
but he went and started using the Dex, adding the value to the product,
and then what he did was six months later when the price went down,
he actually buy back the token.
Now he gets some merit from 1 to 100, a smart investor merit.
So next time when Jeff comes and wanted to launch a DeFi kind of product,
he will be among the 10% of people
who should get better opportunity
to get an airdrop for this kind of thing.
So creating the signal
and then going for API kind of business,
that means market makers,
another kind of people who want the data,
they can go for monthly subscription
and all these things.
So in a nutshell,
building the whole ecosystem
with the DEID protocol
that we are talking about.
Yeah, that's quite comprehensive. Thanks. Appreciate that. Jeff, as we transition into
the discussion of the intersection between AI and crypto, anything you want to kind of connect
the dots here for us in terms of the conversations you're having in
AntID? Yeah, I think one of the things we're seeing, and this is kind of leads us nicely
into the ZKP and the AI section, is there's a lot of things going on in terms of, you know,
civil resistance is still not a fully solved problem, I would say.
And that gets potentially more complicated as we get more complicated AI.
So when Raz is talking about the things he's working on and other people I'm talking to,
it's still there in the back of their minds or at the front of their minds a lot of the time.
Also, on top of that, we've seen an AI plays into this as well.
on top of that, we've seen, and AI plays into this as well, the entire mining of airdrops and just
dumping and moving on and going. So we see huge spikes in price and then a regression to zero,
essentially, and then no value being added. Actually being able to have some sort of reputation,
but not having to give away too much private information on the ZK side, that's important
to people as well.
And so I think everything people are building,
what we're seeing is AI is changing the game across the board on different things
and privacy is back on the table.
People want to talk about privacy.
And so this conversation becomes really important.
It doesn't matter what you're building.
You can be building social fire.
You can be building DeFi, whatever it is.
You can be building DeFi, whatever it is.
It doesn't matter.
It doesn't matter.
Privacy and ZK and resistance to civil attacks and AI influences is on the table for everybody.
Excellent.
Well, let's move the conversation now a little bit more directly on the topic of AI and crypto. For anybody who's
not aware, the conversation that we're going to have today really is, I guess, connected or in
reference to the article published by A16Z, Crypto, earlier in the week or on the 11th. So last week in which some, I guess,
views or the division of how these two will intersect in the future have come together.
And I'm not going to read the whole thing. I'm actually going to try to find the tweet and
pin it up in the billboard for y'all to go ahead and click on. But I'm going to hit on
some of the major categories that are discussed. And then hopefully here with the panel of folks
that we have on stage, we can break them this future, I guess, intersection of technologies.
So the first one is in the identity category. And the vision here is persistent data and context and AI interactions.
And really talking about the ability to have AI systems glean information across LLMs, which currently are in cross-compatible to provide a much richer and holistic view of the information that it's
trained on, but also that perspective that one AI system might have over another because of that
LLM on which it was trained. And the way that it was described
here is that
the blockchain or
zero knowledge proofs in this case
can provide
visibility into this data without revealing
everything, which I think we've talked
a lot about here.
The other one is universal identity
for agents, which this one kind of hits in the
name, right?
We're talking about should agents have a persistent identity that can be referenced, both in terms of knowing who they are, but potentially even applying some sort of like credibility to the work that they do.
Third one here is forward compatibility or forwards, excuse me, compatible proofs of personhood.
I mean, again, proof of personhood we've talked a lot about here.
Proof of humanity, I think zero knowledge KYC, all of these different things are related.
Decentralized physical infrastructure or D-PIN for AI, infrastructure and guardrails for
interaction between agents, and so on and so forth.
So I really just want to start at the top.
And so I'll put it out to Jeff and Moaz here.
Jeff and Moaz here, in terms of this persistent data and context between agents,
what do you think dids or zero-knowledge proofs can provide the necessary infrastructure for privacy to, I guess, exist in a world where, you know, that becomes
more and more important as these LLMs are trying to absorb as much information as they can
to create better agents. I don't know who wants to take this, but feel free to unmute.
I can kick off.
So I was at ETH Prague the other day, right?
And I was fortunate enough to watch Tim Berners-Lee
chatting with Vitalik.
And this is really relevant to this, don't worry.
Tim Berners-Lee was talking about Solid,
his project in Web3, and they have pods, which keeps data in.
And that's quite private data.
You have control over it and you put it out there and you do things.
They're also building their own AI system that works within those pods to access that data and to take actions.
The reason that's important, of course, is because if I want to use an AI properly, and I don't mean just logging on to GPT or Claude or whatever and getting it to help me with some writing or search for something or something, I want it to know about me.
That's when it's really useful.
I want it to know my daily routine.
I want it to know when I walk to the fridge.
I want it to know what I've cooked today.
I want it to know how much I earn, what bills I've got, what bills I haven't paid. I want it to know what I've cooked today. I want it to know how much I earn, what bills I've got, what bills I haven't paid.
I want it to know my education.
The more it knows about me, the better it is, the better it can help my day, the better
it can preemptively do things, the better I can give it the permissions to go and buy
things for me and sell things for me and book things for me.
But to do that well, it needs to know everything about me.
And then we hit problems, right, instantly. Because we hit privacy problems, we hit data
security problems, because that data has to be available all the time. It has to be stored
somewhere for it to be available all the time. And you've got this AI that's going in and leeching
bits of your data to make decisions on your behalf. And at some point, there's some interactivity,
there's some collaboration potentially with that data. And all of that to get the most out of AI opens up an entire
new problem in terms of privacy and data storage and identity. And if we think Web 2 has been
pretty bad at data and privacy, I don't think we've seen anything yet until we see what's going
to happen with we give more and more information to the AI systems so that they can be better for us.
So no answers to this at the moment, other than I do really want to highlight this.
And we spoke about this a lot, Humpty, is this is a problem that is going to end up being all over the news.
I'm absolutely convinced about it because we are giving more and more data
because we want to do it.
We want AIs to work for us
and they require that data.
So two things we have to solve
as a matter of urgency.
And I think this is why people say
the natural home for AI is on chain.
We have to solve persistent data storage,
being safe,
and we have to make sure
that privacy is maintained
as we give away all that
information for AI to do the work for us. Yeah, I'm gonna real quickly, because I feel like this
connects to a lot of the spaces that we've hosted recently, talk about this idea of everything on
chain. If I remember correctly, Jeff, you were very much anti everything on chain,
when we had the discussion about like gaming
and you know this idea of do we need to have fully on chain games so i'm trying to loop this
into previous conversations both because we've had these conversations but also hopefully to
encourage people to go to re-listen though to those because i think they were extremely valuable
pieces of content there that we touched on.
So this idea of like having everything on chain, I actually think that it doesn't need to be, right?
Like does all of this data need to be on chain?
Do these proofs need to be on chain?
Maybe.
But what information does need to be on chain, I think we need to be very careful of because, you know, we've learned anything from blockchains is that they are not private.
They're fully transparent.
And, you know, they're to some degree permanent, right, until we run out of storage in blockchains and we need to prune the blockchain.
blockchains and we need to prune the blockchain. So, you know, hey, let's be honest. But, you know,
So, you know, hey, let's be honest.
for the most part, because of that persistence, you need to be aware that 10 years from now,
that might still be there and your views might have changed, but someone's context to what you've
said might not understand where you're at now. So I think what's important is that we have systems that recognize
the importance of persistency, of, I guess, interoperability, but also, and probably more
importantly, privacy. And I think this is really kind of where ZKPs shine, where you can, and I'll
use it in the context of KYC, because I just talked about this to someone earlier today,
where on OnToWallet, you can create your decentralized identity and go and generate, or should I say, go improve your humanity
by completing some KYC through a third party, have that hash of that identity be preserved
as a zero-knowledge proof, which then can be reused for some kind of repeated proof of humanity anywhere without revealing any private data.
I think this is really kind of how we could potentially provide visibility into parts of ourselves or our daily lives to an agent who wants to have information but doesn't
need to have every single piece of context of that information, just what's necessary in order
to complete a task, right? And I'll use my own perspective as a creator, I have created two or three different agents that consistently are
listening to my content. And they're reengaging that content across different platforms. So
audio only content, short form video content, and written newsletters.
And the beauty of it is that I have the control to some degree because I'm the one that's informing it with where to go and find that information.
So I still can selectively disclose what is it that I'm saying that it should be listening to.
what about, what is it that I'm saying that it should be listening to?
And then from that construct some sort of like intelligible kind of information that can be
repurposed. But if I had no way of like selectively disclosing or choosing when and where,
that's when it starts to get a little scary because it's like, do I need what I'm saying or how I'm singing in the bathtub in the mornings, right? Or when I'm sick and like
I'm ill under the weather and I'm irritable, like do these things need to be contextually
to my point of view when it isn't necessarily like the full picture, right?
So I'll leave it at that because I feel like now I'm just
I'm just kind of repeating the same point.
Stop right, Mas.
Yeah, thank you.
Actually, we also like, when we're talking about how I see it is, I spoke with some people actually who are working on the ZK side and all these things.
Like not long ago, I will not pin the post, but today actually I posted even Jeff, you can find it or I will send it later.
Maybe I will connect you guys later on.
later on, not the same people. First, I was talking like a month ago, people like when a government,
when a citizen goes to a driver license, let's say you did some sort of, you over speeded,
and when you go to police station in the future, when everything almost like everybody,
all the government starts using basically, now in Dubai basically, they are trying to do AI and all this on chain and stuff. You don't need to show the police station for this person, the officer, all the details that's required your health insurance and all these things. You just need to show them the ID card that's required for the police member. Okay, do you have a license? Yes. Do you have enough credits on your the driver points in your license card? Yes. Do we have authorization for license for driving
four-wheel vehicle? Yes. Okay. You overspeed. Okay, fine. We are taking this amount of points from you.
So that's what they need. And the same thing, if you go to an insurance hospital, you show the same
card, and then if it's basically using, the government is using for citizens, this kind of
solution, the doctors or the hospital doesn't need to know all your
things do you have a health insurance okay yes this person holding this card
have health insurance that's all you need to prove it to them you don't have to
show them okay this is the money how much money you have in all this things
the sentiment you got to take a bank loan and all these things you don't have
to submit all these documents for three months and all these things blah blah
is this person having enough credits to get a credit loan for this amount of money is asking a simple question and answer yes or no
simple that's how we basically were talking about and then coming forward to the agents
do you know Humpty I met quite mind-blowing guys a week ago and I have been talking to them back
and forth like almost every evening they actually found agents which actually can come onto the space,
have its own Twitter agent.
I'm not talking about ARXBD kind of agent, but have its own Twitter agent.
Manage it.
Like shitposts like you and me we are doing.
Create content and come to the Twitter space and then say,
if you are the person who trained it, it will mimic your voice, copy it,
copy your style of conversations and start conversing.
And you can actually stop his conversation and interrupt it and do all these different.
So he was actually, I was testing the whole thing.
He was actually, or she was actually talking exactly like a human.
And then the funny part is you can actually use him as a translator. You tell him hey from now on you are a translator for me
I'm trying to speak to an audience from Japan to translate what ontology is doing and you now
basically translate what I'm trying to say in Japanese and he speaks Japanese fluently. So there
are people who are doing these things and this guy is mind blowing from USA and they, he actually found the solution and they, they are going to be executing two weeks later.
I think I will introduce you to them.
Probably it's a good thing.
What, uh, you, you might want to see what they're doing and all these things.
Maybe two weeks later, maybe I will bring my agent with me here and I will talk and I will put the agent here to talk to you.
Yes.
I find it funny.
I forgot to tell you while Jeff, let's say you tell the agent to go home, but don't forget to remind me on my schedules.
And if you actually give your currently access or you get a dashboard, you can connect your Google currentlyly your your currently account your google calendar
and all your memories and all these things and tell exactly the agent what to do and let's say
you are on a twitter space it will actually call you on your phone and tell you hey you need to
stop twitter space and come and do this meeting you have a meeting starting in five minutes
it's it's hilarious how he acts. Yeah, definitely connect us.
It's certainly, personally, I can speak for myself.
I'm very interested in kind of the edge cases or more emergent AI behaviors and kind of solutions that are being built.
kind of solutions that are being built.
I use agents or AI to some degree,
I think quite regularly,
more for my creative endeavors
to have a better grasp
or more perspective about conversations
that are happening around crypto.
And like I said,
for repurposing content
across different platforms and even to analyze and give better visibility into the impact of some of these conversations, you know, on some of these platforms.
So, yeah, I think that that's that's magnificent.
I think that that's magnificent.
I started at least last year early, let's call it February or March,
I started to build out an agent that I was hoping would be
replacing my more mundane tasks by this year.
But it's very difficult,
especially when the data that you're supplying it is quite limited.
But the goal was to have it know the priority of some of my tasks and then some of the intermediate to high-party tasks that would necessarily be more time-consuming than or meaning not as meaningful to try to take
those on maybe meet me halfway and then have me produce the other the other part of it it hasn't
worked how i imagined so anything that can enable that i would gladly give a shot to. Jeff, I see you so eagerly raising your hand.
Yes, buddy.
Yeah, it's just a point, I think.
And interesting listening to you speak
about how you use agents, Humpty.
So I'm going to disagree a little bit on the data
and how much data we need to give agents
to be really useful.
And so I think I absolutely 100% agree with you
when we're doing task-based things,
when we're actually just, you know, it's a functional process. Help me with this creative
endeavor. Help me with this calendar booking. Help me with this Twitter spaces, these different
things. One of the things I think we have to keep in mind is humans are particularly bad at pattern
recognition. We like to think we're really good at it. We're really bad at it. We either don't
see the patterns or we see patterns where there aren't patterns.
With lots of data, and we have lots and lots of data available to us, all of us, because especially
people in crypto, because we spend so much time online and digitally and all those different
things. I think we miss an opportunity if we don't allow big data analytics from ai and the the reason i say that
is you might find that you say you don't need them to know if you're singing in the shower
but actually if your ai agent knows that you've been singing and what song you're singing
maybe they see the pattern and say do you know what when you sing that song you don't have a
great twitter spaces just be aware of that maybe you just need to sit down have a great Twitter spaces. Just be aware of that. Maybe you just need to sit down, have a coffee,
think about what you're doing,
put this different set of music on.
Because when you're singing that song in the shower,
you tend to be a bit more joyful.
You tend to be a bit more vibrant.
And all these different things.
And it's those pattern recognitions, I think,
where I can play a huge role in our day-to-day lives,
which is where my concern of to be really useful, I think we have
to give them a lot more data than we think we do. And so that data has to be good.
Yeah, but I think that's the point that this article makes, right? Is like, how do you
maintain some privacy with the data that you're training these agents with, right? Because it isn't simply about,
should we be giving it more data? I think the answer is absolutely yes. But the reason why I
choose to disclose or select the data that I do choose to share is because I believe that it is sufficient to get it to do some basic task without revealing too much of myself, right?
Which could be used for purposes not intended, right?
Especially if the data is not private beyond that LLM, right?
Or some sort of private AI model that I'm building.
So this also kind of goes back to when social media first launched,
for me at least, using Facebook.
I was very apprehensive about creating an account, first of all.
Secondly, when I did what I shared on social media,
because, and I think some people probably did the complete opposite to their, you know, much to their
own kind of negative impact, because this information is public, and probably even more
so because it has a public forum or arena where anybody can engage with that.
Who's to say that LLMs aren't going to be even more public than how the content you share on social media is seen around the world?
If, say, for instance, I do overshare and overtrain my agents, it can recreate a piece of my persona, which could then potentially have become agentic in that it has its own personality and it can speak for itself.
And it might speak only to the worst part of me or to the most embarrassing part
of me, which is singing in the shower, right? Regardless of how great my Twitter spaces are
when I do. So I think that's the thing is like, how do we share this information and maintain
its privacy to not go beyond its intended purpose? And we can talk about this for a few more minutes,
but I do also want to hit on the other one,
which is identity for agents.
So Jeff, did you have any thoughts on this
as I've rebutted to your rebuttal?
No, no, no.
I agree with everything you just said.
I think it's a difficult problem though.
And I'm not sure exactly how we solve it.
I think combination of on-chain and
off-chain storage is probably the solution with on-chain access though at the end of the day i
think with zero knowledge proofs or zero knowledge systems or even something like zktls and accessing
off-chain data in a zero knowledge way as well is probably the way to go. But yeah, just agree with you on that entirely. Since you brought it up, I don't like leaving terms
undefined. What is ZKTLS, sir? Okay, so that's the ability to go. I can't remember what the TLS
stands for, but it's the ability to go off-chain with an API and check something without checking
all the information. So for instance, one of the things I've been looking at using it for at the moment
is I'm a big gamer.
I've got a great Steam account.
I'm looking at solutions for gamers and on-chain gaming.
The ability to go and check your Steam history, for instance,
off-chain Steam history, the games you've bought, the things you've done.
And ZKTLS allows us to go from on-chain to the Steam API,
have a look at all the things you've done, ask a question about that.
Has this person played X amount of hours of first-person shooters?
Have they spent X amount of money on games, yes or no?
And with nothing given away, we can answer that question in a zero-knowledge way,
which is actually also a great proof of humanity, by the way.
Yeah, and TLS, transport layer security.
I was an IT major once upon a time.
That was one of the first things we learned.
Yeah, it's basically an internet protocol
that enables for secure communications between two systems.
So client server, usually.
So your web browser and some internet server to make sure
that it is encrypted. The ZK piece, just zero knowledge, so an added layer of security with
this new primitive, which is ZKs, ZKPs, excuse me. Now, the next one, I guess we never really talked about the application of ZKPs, but we'll leave it there. We can dig in a little bit later because we only have 15 minutes left.
And I think this one's probably kind of something you and I, Jeff, have been talking about for over a year. So, you know, congrats A16Z Crypto for discovering a conversation we had a year ago.
Z Crypto for discovering a conversation we had a year ago.
Maybe they're listening to our Twitter spaces, who knows.
But basically, this is the idea that
agents would require some universal identity.
Universal, by the way, being the key word here,
you can create an identity system that is
created by OpenAI or an identity that is created by OpenAI
or an identity that's created by Microsoft
and works only within those ecosystems.
How useful is it when you're trying to verify that identity
across incompatible networks or protocols?
Like, it's pretty much useless.
It's just going to be a bunch of gibbers.
DIDs, on the other hand, it's an open web standard, right?
We've talked about this many a times.
W3C open standard that enables anybody to build this universal identity system that is compatible, you know, across a variety of applications because it is agnostic for the most part
in terms of where it's being used.
So if you have a DID on the Microsoft server,
which Microsoft is one of the early contributors
to the W3C,
the Decentralized Identity Foundation for sure,
or Ontology, or mozilla these are
you know verifiable outside of these platforms it's why ons id is something that is able to work
once implemented across a multitude of chains almost almost a dozen chains, where you can carry your
credentials vis-a-vis reputation from one chain to another. It is not siloed to a chain the way
that it would be if you're just using private key identities. So universal identity for agents, let's discuss what are some of the benefits.
I can kick in. I can kick in here. And you see something I'm quite passionate about, actually.
And I think I saw something earlier where Elon was having a bit of an argument with Grok.
He didn't like the reply Grok had given something.
And so he wasn't very happy with his performance.
This is one of the reasons why we need identity, universal identity for agents.
Not just for agents, actually, but for any version of any LLM or AI that's doing anything.
Because what we need to know is if these things are going to be powerful, if they're going to make decisions, if they're going to give information, if they're going to give results, if they're going to guide people.
It's really important that we know what data it was trained on, which version of it it was. And that needs to be verifiable and checkable.
Because if that information or the data training or the performance has changed, if something's
been upgraded, it's absolutely essential that we know which agent made that decision for
us, which agent, which version of that agent, which idea of that agent made a choice.
This could have actually legal implications down the line as well.
And so first and foremost,
it's about what data has been used to make decisions,
what data has been used to make advice,
and an ID linked to data could provide that solution.
That's my quick first answer home to.
I love it.
Moaz, any thoughts about this?
No, this is actually the same people that I was talking about that when I was telling them like if each individual user that you are planning to give, like for me they are
giving one agent and then for you they are giving one agent, how does it going to react
and all these things.
So it was a long book here, the founder told me.'m not that much, like I would say experts in AI.
I like to play around with the agents, but he gave me, he opened a whole book in
front of me, how, how the AI starts learning and all these things from individual
people.
And like Jeff mentioned, an individual AI will be responsible.
So the whole entity will not be responsible, but an individual AI will be
responsible for that particular addition that the particular AI is doing.
Now, if the AI is basically managed by me, now that's me who had been training and whatsoever
this AI is doing. It's my sole responsibility. For example, if it's an AI agent only on Twitter
space, for example, then Twitter space or posting or commenting on somebody like that and sending DMs to people or promotional stuff, that all depends.
So he gave me a whole book of those things.
So I'm not in any way expert in AI.
I just love, I'm fascinated about the technology and all these things and self-studying the whole stuff around the AI things.
Yeah, I would not consider myself an expert either.
I don't expect, I don't think anybody here on this panel is an expert. I would consider myself a fan of the technology and very closely following kind of the progress or kind of the developments that are happening. But even then, I don't think that I'm necessarily the most informed when it comes to AI
and some of its kind of recent changes.
But yeah, I mean, agreed with also what Jeff was saying.
I think this idea of providing some sort of verifiable identity
to any agentic system on the internet is important,
not just for attribution, but for, you know, credibility, I think is important.
Like, my opinion, probably not my opinion, probably fact, I'll let you decide.
A lot of what's wrong with the internet today is that you have a lot of people who hide behind false identities and they claim some truths
that are obviously not. So what is not verifiable is who these people are. What is verifiable,
not easily verifiable, is what these attestations that they're making are.
I know that social media has tried to help in this regard. I know for a while Facebook was, you know, giving you context.
And I think X to some degree still does in terms of like if something is factual based on kind of other conversations or information that the network understands to be true.
But yet, people can still hide behind these avatars that we have and these identities that we've created on social apps.
Again, I like to think of things as progressive, less so as completely revolutionary. just because the new web is kind of introducing these,
or maybe not even new primitives,
but kind of an evolution of things that have existed for a while,
like decentralized identity, cryptographic proofs, and AI agents,
doesn't mean that it's going to behave completely different
from the way the internet does today.
I feel like it's going to only either solve for some of the
previous challenges or maybe even to some degree, worsen some of the problems. And so to the point
I was making before of like, I don't want to disclose everything about myself to people I
don't know on social media, the same, I don't want to disclose everything about myself to,
you know, LLMs that I have no knowledge of how they're going to reuse my information.
The same thing here in terms of like the, you know, kind of accountability, attribution, and credibility.
When we think about like what's happening on the internet today in the Web 2 form of it, which is the social web,
in the Web 2 form of it, which is the social web,
I kind of would think, well, as it becomes more agentic
and AIs become a part of the internet as much as we are,
if not over-proliferating the internet,
where there will be more bots and agents than humans,
I need to even more so know if the people or the accounts are, you know, human,
if they are, and even probably less so if they're human, if they're reputable. Like,
are they out there just being a troll? Great, let's just recognize them as a troll account.
A troll? Great, let's just recognize them as a troll account. If not, if they are out there
expounding truths, are they usually right? Are they just throwing darts on a wall that's pitch
black and sometimes right? Or are they proven to use some scientific method to more accurately
portray reality? These things are important, I think, to most of us so that we can have a
general idea of what is true that we read and interact with online and eventually on-chain.
Go for it, Jeff. Yeah, I love that idea of being reputable, Humpty. I think it's really important.
So when we talk about identity and when we talk about reputation and proof of humanity and all these things, it's often because we're trying to
combat something. It's because we're trying to solve a problem. But think about the opportunity
here as well. Ignore solving a problem. Think about the opportunity. We might want agents to
go out and act on our behalf, right? Think about what M Waz was saying earlier about, you know, somebody might back a project, do some stuff, get an airdrop.
They might sell their airdrop, but they carry on acting
with good faith for that project.
They buy back and all these things.
Actually, moving forward, that's going to be less and less likely
to be a human.
Proof of humanity, whilst important, isn't going to be the only question we ask here. Proof of
reputability may be the question. So is my agent behaving in a good way? We can acknowledge that
it's an AI agent, but we can only look at if it's behaving in the way that we want it to behave,
if it's behaving in a way that brings value to a system, if it has an identity linked to it, and potentially an identity linked to my identity as well,
if we wanted to go that far. Actually, giving these agents a universal, verifiable identity
that can be linked to the way they behave opens up so many opportunities. It can go off and it
can buy music from me. It can stream music for me. It can take part in DeFi.
It can add rock farm for me in a positive way
and act in a good way on protocols
and be rewarded for that.
And likewise, somebody who has an agent
that just goes out to extract value,
if it's got a reputation and identity attached to it,
we can see how that's behaved as well.
So actually, giving agents an identity and having them behave in a certain way and look
at their reputation and how they've behaved is a huge opportunity for projects and people
alike in reality.
Yep, 100%.
Well, we've reached the top of the hour.
I feel like time flew by.
And that's how you know it was a good conversation.
Sometimes, honestly, I wish this was longer than an hour,
but sometimes, to be honest,
an hour is just enough to get us going and get spicy
and definitely open up an opportunity for discussion
the following week,
which we've left the gaming chat on hold now for two weeks,
but I feel like we've kind of,
at least for me, I'll speak for myself, Jeff,
we kind of were ignoring the base of what it is
that we talk about regularly here,
which is privacy and kind of DID and ZKPs.
And during that time,
a lot of really cool stuff happened
in that space that is hard to ignore.
So I want to thank you, Jeff,
obviously for, like myself,
being out there exploring conversations,
bringing them to each other
and, you know, kind of having
deep conversations about it.
And Moaz now for coming back
and having this conversation with us.
I think this is the third time, hopefully not the last time. So I feel like you've enriched in the
conversation because you're deep into developing in this space as well. So as we sign off, I think
next week, I'd like to continue this conversation because I know there was a few folks who missed
the call this week. And maybe we can go deeper into the actual application of these primitives like ZKPs and DIDS.
And again, kind of like what we did maybe a week or two ago,
explore the actual development of these, you know, in public.
So, Moaz, let us know where we can find you and Jeff after that.
So those people who doesn't know me, they can find me on Twitter.
I'm building.
Hopefully when I'm finished with the first thing that people can use, I'll be talking
about it.
For now, I only talk to very close people that I trust and all these things to create
the community of the whole thing.
So I don't show I don't, I don't want to show that the whole thing out, but Jeff and
Humpty, I send you both the tweet.
The reason I didn't want to pin the tweet was after going through talking to them
last whole week, they're trying to have an NFT.
So I didn't want to promote the NFT and all these things because I'm not even
mentioning the game.
So the kind of utilities have having and all these things,
after this call, I will actually try to connect you guys with them.
I think it's quite interesting, Jeff, with you guys and them.
I think that there's some kind of synergy.
I don't know what kind of synergy,
but I'm sure there's a synergy with what they're doing
and all these things with all the different agents
and all these things.
Maybe, I don't know, maybe Jeff can have an online girlfriend.
Don't let my wife say that.
That's fired.
No, I'm serious. It was hilarious. Can you imagine, Jeff? You tell this one, hey, can
you find me a girlfriend? And then this guy is trolling you. He's your agent, and he is
basically setting up and saying, okay, you think you are too down because you did this today?
I will set up you with somebody and then somebody calls you.
It's crazy.
It's more than that.
I think when you see it, you will know better than me why I posted about that.
You know, that sounds just like the evolution of 1-900 numbers, but go for it, Jeff.
You're just concerned they're going to need an agent to find me a divorce lawyer here, Marz.
That's the problem.
But no, hopefully, look, hopefully we can, Humpty, you're going to send me some details about the Germany stuff, Marz.
Maybe we can even hook up in Germany and catch up in person.
We can take a great photograph and we can compare our glorious hair together as well.
Great conversation, Humpty. Great conversation, Maz. Really, really enjoyed it.
So looking forward to next week.
Awesome. Thank you, everybody.
Make sure you follow Ontology, which is the host account here on Ontology Network.
And you should be able to find on Ont.io all the links to the Telegram and Discord
if you want to join the community.
Until then, we'll see you again next week.
We'll be back same time, same channel.
Have a great weekend, everybody. Thank you.
