Transcription
MIT study that said people use chat to
get dumber like we don't think an AI is
going to replace a human but you got to
start using AI the one thing that's
powerful about human ingenuity which is
also why I think AI won't replace us
there is a kind of sheer will that
entrepreneurs and builders and just
human ingenuity has the kind of
ingenuity that had to survive crisises
and wars and all these problems are the
same ingenuity that we have to basically
build companies and build businesses and
you know see world opportunities that
others don't
>> 2024 was kind of like solarreneurs these
are preed tier one speed act startup
they're asking the question should we
use AI and then 2025 Five was how should
we use AI? What we're seeing this year
across all those segments in industry is
how do we operationalize AI across every
segment of our business.
>> We're all talking about you know one
person sort of you know companies
because an AI agent can do so much work
for you tokenize or die. If you don't
have a token strategy if you're not
tokenized you will lose against the
company that is tokenizing. There are
all these technological shifts that
happen that basically you have to be
ready for to address these new markets.
>> You should be testing multiple things
internally. There's no reason why you
can't do that. If you're leveraging
these agent coding tools like cursor or
or curo or or cloud code, you can have
multiple windows opening running
simultaneously running different agent
workloads and getting to a point where
you don't even just have a PC. You
actually have to have a production ready
uh tool deployed in the cloud built to
scale for nothing. So you can get
credits to different cloud providers
like AWS activate. So there's no excuse
not to be able to go to market quickly.
>> The ability to become an expert in
everything is so much more available
than ever before. Also fail fast.
>> We wanted to take them away from the
chatbt search experience that experience
itself limits you. that you go into a
website experiencing like search you
immediately start thinking what should I
ask it that's not the way that you're
supposed to frame it oh it can actually
respond in this manner and then when you
begin from that level the imagination
opens up
>> the difference between somebody that
succeeds as a founder or a group of
founders or somebody that doesn't are
the ones that are committed to that
learning experience then dive in deep
and you understand you know how to use
these tools how not to use these tools
what their limits are what these limits
are and then the light bulb moment
happens
>> when capital meets infrastructure. We
have Yatsi from Animoka Brands and Ian
Holtz from AWS.
This is probably one of the most
important conversation you will be
watching if you're a builder or an
investor that is interested in tech.
>> Hello guys.
>> Hi.
>> Thank you for having us.
>> How are you feeling?
>> There's a lot of energy.
>> Yeah.
>> Yeah,
>> it's a lot of energy. I I always you
know if the markets are down which they
are a little bit in crypto shall we say
>> I tell everyone just come to a crypto
conference and re-energize and you know
I would say true to form consensus has
been that
>> see that that's the most interesting
part because every time when I go to a
conference we talk to builders and
everyone's so excited
>> that's right
>> but the sentiment and you know re among
retail investors it feels a little
different
>> yeah but I think we need to distinguish
between builders and investors as well
and particularly retail investors who
are really oper operating a little bit
on fear because they have uncertainty.
They don't understand what's going on.
And it's not just about what's happening
in crypto. It's all macro, right? You
know, what's happening with Iran, you
know, what's happening with Walsh? Is it
is it good for the economy or not? You
know, all these things that happen all
at the same time. We just don't know,
right? And that uncertainty is
unsettling for everything. It's not just
it's not just uh not just crypto, right?
Uh but the builders, they see a future,
right? If you're a builder, doesn't
matter if it's AI or crypto or whatever,
they're kind of all sort of intersecting
anyway. You're you're not here for
today. You're here for tomorrow. And if
you're excited about tomorrow, this is
great. So come to conferences like this
to see tomorrow. Right.
>> And you guys are building towards
tomorrow. So basically, you decide what
tomorrow looks like.
>> That that's my interpretation, right?
>> That's a bit bold.
>> Yeah. Yeah. Well, I would just say um
>> uh at AWS, we provide the
infrastructure. Yeah.
>> Uh for the builders. They're the ones
that are innovating. Yeah. So from our
perspective uh we're democratizing
access to technology. The builders of
today that are innovating uh for
tomorrow uh whether those are students
um or you know tier one invested VC
startups from preede to series B they
have access to the very same tools.
>> Yes.
>> And so you know what I also noticed um
you know conferences like this uh
outside of the sentiment that we see on
the media is uh you know the convergence
is real right. the convergence that
we're seeing um you know with whether
it's robotics or you know AI coding
accelerating workloads uh is happening
here um and so outside of trading
markets um that's the energy that that
we're seeing so
>> and I think just to add to that you know
when you build the tools for that and it
makes it easier for innovators to come
in they can ideulate and focus on
creativity rather than just the tech I
mean you know I was there for the early
days of the internet right and I
remember to set up a web server I had to
buy my own box I had to literally build
it uh and of Whereas today I can do
everything virtually. I just do it on a
dashboard and voila I have a server and
the abstraction layer goes even higher
and higher between applications between
services. And this is the thing about
blockchain. Blockchain is another one of
those layers that abstracts essentially
the security layer of anything that has
value essentially whether it's financial
whether it's asset holding you know you
know one of the big topics last year was
prediction markets like poly market. I
don't think a lot of people know that
poly market was built on blockchain. I
don't think more people care. However,
Poly Market would not have been able to
be built in the speed and the manner in
which it has if it wasn't for
blockchain. Because think about what it
would take to build an infrastructure
that can handle your money, that can put
a bet on the millions of dollars of
size, that can do all that type of
stuff. You'd have to sort of hire
someone that basically can build like a
quasi mini bank, which would cost not
millions, but probably hundreds of
millions of dollars infrastructure. That
infrastructure is not just available for
poly market is available from the 16 to
14 year old to the 70 or 80 year old and
they can all innovate on top of it. add
vibe coding to it and you know my mom
can start coding financial implications
if she wanted to because the hard part
is taken care of right and so I think
we're entering a world that's going to
be much more creative which is why we're
also seeing a lot more startups and
entrepreneurs that's actually
accelerating it's not declining um I
remember being a startup uh an
entrepreneur in the '90s
>> it's a long time ago um meant you were
unemployed that's really what it was uh
nobody would hire you so you start your
own company right um and now it's kind
of expected almost right you if you if
you if you go to your parents and you
say especially Chinese parents or maybe
Asian parents generally and you say hey
you know uh I'm going to do my own
startup
>> try that 20 years ago right um and now
it's like
>> okay go ahead
>> that's go ahead that's cool right um so
yeah it's changed a lot
>> do you think capital decides where
technology goes or is it the other way
around
>> you know I I don't think uh those are
opposing
>> um I think they compound each other so
if you start thinking about investment
without infrastructure, that's
speculation. You start talking about uh
infrastructure without investments,
that's um you know unused compute. So um
they're not opposing forces, they're
compounding forces in my view.
>> Yeah. And I think also just to add on
that I mean the one thing that's
powerful about human ingenuity, which is
also why I think AI won't replace us,
right? Even when we're talking about AGI
and all that kind of stuff which it's a
different topic alto together uh is that
there is a kind of sheer will that
entrepreneurs and builders and just
human ingenuity has uh you know the kind
of ingenuity that had us survive
crisises and wars and all these problems
are the same ingenuity that we have to
basically build companies and build and
businesses and you know see world
opportunities that others don't right
and then they create this kind of impact
and now because the cost of that
deployment has gone lower and lower and
lower I mean we're all talking about you
know one person sort of you know
companies uh because an AI agent can do
so much work for you um and it's reduced
over time I mean you know you had to
hire so many more people just to run a
website and now you basically just do it
yourself for instance as an example that
type of stuff makes it just possible to
just grow and the cost isn't that high
anymore right
>> okay
>> um so so I think the you know when we
look at for inance our investment side
because we have a lot of them right
>> the uh the sort of seed investing
startup phase was often I have an idea
and this is what I need to get my idea
going. Okay, fine. Here's the equity.
Here's I do but he didn't have a
product. Now seed is here's my product.
It's built. Uh it doesn't really quite
have maybe product market fit, but maybe
you can help me with this. It's
completely different. We're basically
evolved to really the stage in investing
of a seed company is much more like a
series Aish type company or pre- series
A say even 10 years ago, right? Um so
the cost is low and the risk is much
lower as well. People are talking about
AI agents, but I want to know how far
away from, you know, people actually
building companies with AI agent. How
far are we?
>> What we're seeing from builders is as
fast as it's moved, uh, there's still a
gap, right? So, there's a there's a big
gap between an agent uh, buying an
airplane ticket versus an agent for
sure. Right. And so, we're clearly in
category one. M
>> um uh there's a ways to go to get to
category 2. Uh we're seeing glimpses of
it.
>> Um but we're not quite there yet.
>> Is it is it easier for let's say if I'm
a solopreneur, it's easier for me to you
know try out AI agents and all these
different tools.
>> Absolutely.
>> But when we talk about big companies,
they're very careful when it comes to
>> Well, I think there's a different
there's a different perspective as well.
When we come about big companies,
there's a lot of private data that they
have to protect. So it's about policy. I
mean, you know, uh I don't think many
corporations will be happy to install
open claw right into their local
servers. I mean, in their local
infrastructure, that would be a problem
because they're going to do and see
stuff that is like what what is all this
about. However, and if you're a solo
entrepreneur or a small office, the
risks are much lower and you can
experiment around that. Um but I think
what you saw with open claw, it kind of
opened up the possibility to say, wait,
hold on, it can kind of do this stuff.
And to your point, it's uh you know
again because we don't think an AI is
going to replace a human, but the
instrumentation is going to be happening
through an AI agent who can do all this
stuff for you as well, right? So you
don't have to do all the minutiae, set
up the GitHub, set up the
infrastructure, even write down the sort
of draft email, that kind of stuff. It
can do 90% of that or 95% of that and
then you go in at the final stages and
edit and review. So you have to
understand it, right? It's not going to
it's going to write well for you, but
you still have to know how to write,
right? you don't know how to write then
you know whether it's code or whether
it's English or whether it's Chinese if
you don't know how to do that
>> it's not going to help you right so you
still need the foundations but when you
have that it can be a really really
powerful agent um the other thing of
course is uh what I find really powerful
what you know what we have with animal
minds which is our own sort of AI agent
is it's a way in which an agent can
neutally organize essentially
collaborative workspaces in really um a
sort of apolitical way because you know
explain that
>> well So if an AI agent basically gives
you an output or makes a suggestion or
even says it's a dumb idea, you won't be
offended quite the same way, right?
Okay.
>> But it's very different if a coworker
you work together with basically says
that is stupid. I'm not saying that an
AI agent should say that, but the
criticism is very different. And so they
actually great arbiters, not because
they're the ones that have the final
judgment, but you will look at it not in
the same way, right? And the other thing
is that um because you know AI agents
like this have persistent memory, they
can basically recollect stuff for you on
instruction that you might forget
because you know um we have one side
incredible creativity and that kind of
consciousness that AI doesn't have but
that same creativity because we have
feelings and we have emotions can impact
us in ways that an AI will not right and
so what we are seeing right now is that
an AI agent is going to be in every part
of your workspace to assist and support
you not to replace you and then
basically to accelerate your work
effort. Uh and and also in in family
planning, in sort of life schedules,
calendar, booking tickets, right? I
mean, that's going to do a lot, right?
As you said, um
>> you know, I think, you know, we're in
Hong Kong, we get spoiled in Hong Kong a
little bit because we have a lot of
help.
>> Most of the world does not, right? So,
so most of the world could take good use
of an agent like that.
>> Yeah.
>> I would just add to I mean, you asked
about solarreneurs. Yeah. touched on um
you know data sovereignty and you know
the skepticism around trust related to
you know open paw and enterprise um I
will tell you from an infrastructure
perspective if you look back 2 years in
in 2024 versus 2025 versus this year um
we're definitely seeing shifts
happening. So you know 2024 was kind of
like and this is solarreneurs these are
um you know preede tier ones backed
startups these are enterprises across
the spectrum across the segment across
geos. asking the question, should we use
AI, right? In 2024, right? And then 2025
was not should we use AI, how should we
use AI, right? And so what we're seeing
this year across all those segments and
industry um is how do we operationalize
AI across every segment of our business?
>> And so um yeah, these things are are
happening. Um there's still a gap, like
I said, between, you know, booking the
airplane tickets and and running your
business. Uh but that reality is fast
approaching. you've seen a lot of
builders. What are some, you know,
things and trends that you've seen that
excite you the most?
>> You know, I cover Agenta coding tools.
>> Okay.
>> Yeah. So, at AWS, we have a flagship
tool called Kira. Um, but we also
support, you know, anthropic cloud code
uh natively within Bedrock. Um, you
know, I work with a lot of startups that
are supporting a variety of other agent
coding tools. So what I've seen um you
know as as a developer in the industry
for uh for a very long time um is we're
seeing a 10x output in in um in from
developers and so you I'm working with
senior level developers that had started
tinkering with these tools a few years
ago wasn't really working was creating
all this slop to ones that are fully
embracing it and they're actually not
writing any code. So you start talking
about you know two week sprints being
conducted in a day you know there's
terminology around that around you know
conversion between um you know sprints
to bolts and so what took teams of
people um you know to develop product in
three or six months is now being done in
weeks um and that's again across across
um segments geos and and enterprises to
to startups so um that's what I'm seeing
is uh really an acceleration by
leveraging uh these agent coding tools
>> I mean we're particularly excited about
how how the entire framing of the web is
basically changing into one where Maybe
this potentially could be one of the
biggest challenges for at least the
incumbent platforms. Uh and and the
reason why is that because the biggest
business model for them has been
advertising.
>> Okay.
>> And when you think about advertising
basically in energetic world like we're
moving into sort of you know web 3 was
the foundation that layer where you
could have ownership through blockchain
and basically I don't know if you want
to call it web 4 but essentially the
agentic web effectively allows you to
pull all this information. We're still
used to going somewhere to pull
information and we're not used to
getting this information pushed to us
because there's a lot of slot as in even
an AI recommendation engine typically
will send you stuff that you don't want.
So you don't really care about it. you
want to go to the destination and if you
look at chat GPT as the sort of OG of
this sort of you know LLM moment this
sort of chat GPT moment as they say the
interface was search
>> it wasn't a special innovation it wasn't
like an agent talking to you
>> uh it was basically just a better search
and many many companies are already
saying wait hold on a second I'm not
actually getting I'm getting less clicks
to my website because now if I'm going
to Google Gemini is giving you
everything you need to know I don't have
to hop into that next stage right so
there's going to be this whole movement
around okay how do I get my fair share
you know YouTube had to have content ID
with the creators there's going to be a
version around that there's going to be
a whole movement around you know it's
already happening with MCP servers right
so we're moving from HTTP to MCP right
the user is getting everything delivered
to it through the agent right uh whether
it's the messenger whether it's web
application that you're using um and the
entire framing of the web is going to
change from the way that we're
experiencing it uh more efficiently more
faster but also as a result of that this
This is why we think the token economies
become so powerful because how are all
of those sort of MCP sort of uh service
are going to sort of exchange value with
each other. It's going to be through a
token infrastructure which is basically
one that is something that you can
verifiably prove but that's a stable
coin or another token that's kind of
beside the point. It's like oh I know
this is real, right? Uh and then by the
way this is also why digital identity
becomes so critically important. Uh,
which is why we're building rails around
that too because I need to know whether
this agent is verified. That's right.
>> Right. We're going to have more agents
transacting than humans. And I don't
mean just money. I mean like taking
data, doing that kind of stuff, right?
How do I know that he's authorized to do
so? You need an identity layer that, you
know, that has a reputation stack that
could be entirely digital. That could be
just an agent, right? Uh, and we're
really excited about that because
through that infrastructure, maybe we
can now establish the framing that we
were hoping for with web 3, better
sharing of that value.
>> Mhm.
>> Cuz right now it's not transparent,
right? Um, you know, Google controls
essentially much of that supply, right?
Effectively directing advertising, you
have to pay them all the money. Uh, what
happens if I now can free sort of token
economy? Now you have also third party
people who can actually also bet into
that, so to speak, say, hey, you know
what? I think this website might be
popular with the data they're creating.
I can actually be an investor in the
asset class of that data. Right? Data is
actually and has always been a commodity
but it wasn't possible to do so until
tokenization.
>> You know um RWAS is an interesting
comment topic because everyone says a
real world assets is it real estate? Is
it stocks? Yeah. But you know for our
children when we tokenize video game
items as NFTs that was real world to
them right it's super you know data is
basically I would say an RWBA it's just
never been framed in this way because
you know you know when we talk about
sort of intellectual property rights and
taking royalties and turning that into a
tokenized yield product we call that an
RWA but what is IP it's entirely virtual
right so really I think RWA is just a
fancy way to try to connect the physical
to the digital
But to the human mind, it's all real.
Anyway,
>> I think in one of your interviews, you
talk about in the future, every company
will launch your token. Is that correct?
>> What I said was tokenize or die,
>> right? So, what I meant by that was that
if you don't have a token strategy, if
you're not tokenized, you will lose
against the company that is tokenizing.
Right?
>> Explain that.
>> Well, so think about the early days of
the internet when we had this thing
called digital literacy. You lost to the
company that went online. If you're a
media company, if you're newspaper and
the your newspaper competitor went
online, you lost to that guy, right? If
you were selling books and there was an
online book seller, you lost to the guy
who was an online books clearly, right?
Um, you know, predecessor here, right?
So, yeah, the point being that there are
all these technological shifts that
happen that basically you have to be
ready for to address these new markets.
If you were not on on the app store and
your competitor who was in the app
store, he would win against you. If you
were on social media and your competitor
was not, you will win against, you know,
it's a classic thing. Yeah.
>> So for tokenization, it's the same. But
for tokenization, it's not only because
agents can read your asset. That's a big
one. Or maybe that's a youth asset class
that people like.
>> It's also
>> the fact that you have now new financial
instrumentations available to you that
make it possible to to leverage your
assets, right? Which if you can leverage
your assets that you couldn't leverage
before and your competitor can't do
that, game over for your competitor,
right? So that's what I mean by tokenize
or die. Now if you happen to be
fortunate enough to be in an industry
where tokenization doesn't matter
because nobody tokenizes it's all same.
>> But my point is that if you can figure
out the token strategy which we think
you will be able to just like how you'll
figure out your internet strategy back
in the day right then um you will you
will you will win and every company has
to think about how do I tokenize because
if you don't tokenize you'll be left out
of this agentic web. If someone is
watching this episode and they really
want to start their own thing, you know,
they're very inspired by what you guys
are saying, what are some really cool
ideas, you know, you can come up with or
any suggestions to whoever that's
watching?
>> It's a great question. Uh, I talked to a
lot of founders and,
you know, if we're talking about web 3
in general, um, we're definitely seeing
it not being limited to, uh, you know,
crypto or gaming. It's impacting
everything from supply chain management
to database analytics through um you
know zero knowledge proof and you know
mining data that's not accessible
before. So really the sky's is the
limit. What the advice that I do give to
founders is you know pick the problem
>> uh pick something that you're passionate
about make sure that the economics work
uh and find product market fit right so
you know I've even failed as a founder
where I I built something that I thought
was incredibly compelling it was also
innovative it worked um but it didn't
have that product market fit and the
economics didn't work mostly so you know
we were able one of my early stage uh uh
startups where you know we're still able
to onboard you know hundreds of
thousands of users uh but the economics
didn't make sense So um you know just
because you can create a product doesn't
mean you should create a product. So um
you know I won't I won't uh try to guide
them into a particular segment or or an
idea because I think the sky's is the
limit with the tools that they have
today with you democratizing access to
technology uh the agentic workflows that
are coming. Um it's really uh you know a
oneperson show with almost nothing um
you know can can build a working product
and find that product market fit and um
you know I myself like all of the other
passionate builders around the world you
should be testing multiple things
concurrently there's no reason why you
can't do that if you're leveraging these
agentic coding tools like like curser or
or curo or or cloud code you can have
multiple windows opening running
simultaneously running different agent
workloads and getting to a point where
you don't even just have a PC you
actually have a production ready uh tool
deployed in the cloud built to scale for
nothing right so you can get credits
through different cloud providers like
AWS activate so there's no excuse not to
be able to go to market quickly
>> you guys are giving out credits right
>> we are yeah
>> tell tell us about it
>> yeah so uh we have uh we offer uh out of
the box up to up to 100k in credits
there's exceptions up to 200k and beyond
for jai use cases so um yeah you can
find credits for AWS at uh startupsawws
it's a it's our flagship ship program
for founders and it's not just about
credits um you know the credit part's
important because we want to eliminate
barriers to infrastructure
>> uh but it's also providing uh go to
market support um it's also providing
technical support so um again uh our our
goal is to provide the tools and provide
you know democratization and to lift
those barriers on infrastructure so
founders can get to building and scaling
>> yeah I mean I think the one thing I
would say to founders uh I mean there's
so much right um rather than you talk
about a specific category is you just
have to be excited about learning all
the latest new technologies and just get
into it. If you just if you don't have a
learning mind frame,
>> it doesn't matter what it is, right?
So you know it sounds very trope you got
to be curious
but it is true because you know you
might have a great idea and you know you
want to solve a problem but if you lack
the curiosity to go dive deep into
technology well by the way you don't
have to be a technologist to understand
it but you just have to dive deep into
it and kind of become and that's the
thing the ability to become an expert in
everything is so much more available
than ever before
>> right um so
>> and you can also fail fast
>> and you can fail fast exactly and you
can just dive into this and then when
you use the tool tools and you're in
that ecosystem then you go oh that's
possible and this is possible and that's
the idea right um you know I think the
problem a lot of it is uh I sort of I
say it's kind of like the the sort of
the international China thing which is
like oh China has a billion users back
in the day right and they're like if I
if I if I sell shoes there it's going to
do really well right
>> you just need 10%
>> yeah I just need like well yeah even 1%
yeah my goodness it's billion Chinese so
let's do that let's just roll it out I
just go do and then turns out it's much
harder because guess what
>> they had to go to China Maybe I have to
speak the language, may I have to learn
stuff, uh get a bank account,
>> get a bank account, go go go and meet
all the vendors and suddenly, you know,
through that process of that whole step
>> after he's done that for, you know, a
year or two or 6 month or whatever, then
they become an expert. And then they
identify, oh, this is how I do this and
this is how I sort of navigate this.
Now, because of AI and because of
blockchain and web 3,
>> you can do this all in your own bedroom,
>> right? you don't have to travel anywhere
>> but you do have to go in. So you know um
we meet founders who say they use AI and
we see this in the business plan and
then it's very easy when you talk to
them a little bit actually how much AI
are they really using right are you
using AI to get chat GPT to basically
write your business plan or use notebook
LLM from Google which is a fascinating
product by the way or are you actually
going deeper with it how you optimizing
those workflows matter
>> but matters greatly okay because you
know um to me you know using chat GPT is
what everyone else can do it's like I'm
great at Google search good for you but
it's not something that is not something
that gives you a particular competitive
advantage because everyone else can do
that but how are you optimizing your
workflows uh is gives you the insight ah
you get it or this is what you're doing
>> and it's even away from the jargon like
it doesn't matter which LLM you're using
per se right it's more about what are
you trying to optimize the LLM will come
to it when you know what you're trying
to optimize for right
>> uh and and that's why you know you got
to use some kind of framework right you
got to start using AI um because you
could be using AI just for a business
solution. Nobody said that you had to
program everything from scratch. If you
know what it could do,
>> then you can imagine it. Then you go
into the rabbit hole of saying, "How do
I solve that?" Right? Um and and again
with web 3 as well, we've had so many I
mean back in the early days so many
companies said, "Oh, I want to tokenize.
I want to tokenize." Have you ever
bought a token? No.
>> Okay. Do you have a wallet? Uh no.
Right. It's impossible to have a real
conversation then because they they um
they see the prospect but they don't
understand it. So you have to you have
to do the work.
>> Let's say you're a person that is really
interested in this new AI wave and they
want to build a business around it. But
you know the analogy I would use is if
you can't build a phone maybe you can
build a phone case. You can start a
company that sells phone case. or if you
can't even do that, can you open a shop
that helps people put on the protector?
Like there are so many different levels
when it comes to the industry.
>> So I I know that analogy. It's a niche,
right? So instead of creating a segment,
you're creating a supplier to a
particular niche. But um you know with
with AI, I think it I mean it really
could be anything. I mean it's kind of a
non-answer, but it could be could be
database and analytics. It could be uh
scalability. It could be helping
identify um you know uh customer
segments. It could be SC. I mean really
the the sky's is the limit. I don't
think there's one single answer. I mean
maybe Yat you have a different idea. I
mean I think you're going to lean on to
tokenization obviously but
>> uh
>> yeah. So I don't think there's I mean in
in um you know the the movement that
we're seeing in the AI impact I don't
think there's there's a single industry
that outshines the others that can be
impacted or or or um capitalized on.
>> Yeah. So on the token infrastructure the
re the way we think about the token
infrastructure it lays the foundation
for agentic AIS right so so take that
aside there's a there's a model around
that I think most people are going to be
approaching it from the agent or AI side
where the token just becomes
infrastructure to support them on the
transaction on the system and whatever
you know to make it optimized in that
manner right um but on the AI side let's
say you're not technical at all you know
you can basically just start using one
of those agent frameworks right and
they're getting so much easier so you
know we built one called Animal
Commands. And again, you don't have to
use ours, but this one activates over
email. And we decided to do it over
email because it was an interface that
was easier for you to understand for
most people cuz they're still using
email in our work.
>> And by activating it over email, I don't
go to a website. Uh we wanted to take
them away from the chat GPT search
experience because that experience
itself limits you, right? But if you so
so if you go into a website and you're
experiencing like search,
>> you immediately start thinking what
should I ask it when that's not the way
that you're supposed to frame it, right?
>> Um and then maybe there's prompts and so
and over time they start reacting to it
and then you start to imagine oh it can
do this I didn't know this right and so
many people are surprised oh it can
actually respond in this manner um and
and do that and then when you begin from
that level the imagination opens up.
It's not about do I choose cursor uh or
do I want an abstraction of that like
lovable or like whatever like these
toolings right I mean there's great amaz
you know 11 labs with voice right or you
know you know with Gemini the amazing
thing that they're doing on like all
sorts of things you can do from a
tooling standpoint um those are tools
>> cure is a good one too
>> yeah cure is cure is great
exactly of course uh but what I'm saying
is that that requires a certain level of
knowledge already right so it's like
it's like if you know what kind of car
you want to
because you've basically studied
different cars and you've gone into the
rabbit hole, then you'll find the parts.
Do I need Plexiglass? Do I need this?
You know, which engine do I want to use?
You know, like those those things come
when you know what you want.
>> But you got to start somewhere. Like,
you know, I don't know any car builder
that doesn't drive.
>> Yeah.
>> I mean, right. Right. And you have to
have that passion around. The nice thing
about AI, it's so ubiquitous, you know.
Um, was it maybe 5 years ago or 6 years
ago that you know if you asked their
children what was the top job you wanted
it was like YouTuber or something or you
know some social media influencer
>> a great job by the way
>> right yeah yes of course right why was
that right because it was something that
they engaged in they were passionate
they could see it
>> and similarly you know there's no rule
book you know about it right but then
you start saying oh okay so which
platform what are they doing you dive
deeper into this area and then suddenly
you find your areas of expertise that
you can emerge from that and sort of
optimize. But it's the same thing,
right?
>> Yeah, that's making a a great point
about about diving in deep and and I
mean there's no greater time to do it
than now, obviously. Um, but you you I
mean, as you dive deep, you learn the
the limits of these tools. Um, but and
you learn you eventually learn how to
use these tools. Uh, but also you learn
how not to use these tools,
>> right? And so, um, you know, regardless
of the tool that you use, you know, I
truly am tool agnostic as long as you
are using a tool, right? And so, um, you
know, I'm building all the time. I'm
failing all the time. Uh, and and I'm
learning how to use them, how not to use
them. I'm deploying um, you know,
products that are overengineered, right,
that are not um, cost optimized. And
then what I'll do is I'll create another
project. I'll reference that product and
I'll say, okay, iterate on that. These
are the lessons they learned that. And
so that's that's not something that that
was taught. That was something that I
learned as I dove in deep, right? And so
the more that you and you know the
difference between somebody that
succeeds as a founder or a group of
founders or somebody that doesn't are
the ones that are committed to that
learning experience and and diving in
deep, learning and iterating potentially
on multiple tools simultaneously. And
then once you dive in deep and you
understand, you know, how to use these
tools, how not to use these tools, what
their limits are, what these limits are
at the same time concurrently, these
tools are getting better and better. And
that's when you start to figure out this
is the path. That's when you also start
to figure out how does this um uh how
does this align with what my what I'm
passionate about? How does this also
align with what my resources are, what
my networks is like? And then that's
when sort of you know hopefully if you
follow that process that's when the the
light bulb moment happens. Um and then
you're off to the races.
>> The light bulb moment comes when you're
solving a problem.
>> Yeah.
>> Right. For even when you talk about I'm
hitting the limits of this.
>> Well, how do I work around it? That's
problem solving.
>> Yeah.
>> Right. Is it another tool? Is it a
barrier? Is there a workaround? How do I
do it? And this is also an element of
grit. I might not know how to do this
now, but I'll figure out how to overcome
this particular mountain so that I get
there. Right. And frankly, you know, if
you want to go slightly philosophical,
that's how we will have an edge over AI.
>> Oh, I like that.
>> Yeah. Because, you know, this learning
mindset and this ability to problem
solve on the fly and being able to sort
of deal with uncertainty
>> is not something that an AI can do
because it's not trained for that.
that's trained on existing models and it
will give you good choices but the
ultimate decision framing is yours and
and if you don't if you delegate
everything to AI you know there was this
MIT study that said people use catch to
get dumber right um I mean you know
typical headline you know it was a
certain number of people not all right I
guess they they they just looked at and
I can but I can see how that happens you
know if you're basically just giving the
essay if you don't know how to sort of
write and you give the essay to chatb to
write an essay for you
>> you don't know how to write Yeah,
>> right. You just you don't know the
principles of it. It just wrote it for
you.
>> Um and it may be good or maybe bad. Uh
but the point is is that you've gained
nothing from it,
>> right? You've not iterated yourself,
right? And so the ability for you to
then navigate that and know how to make
it better goes away. So this idea of
schooling through the hard work whether
it's math or or or even sort of you know
the scientific approach these are just
methods for you to basically iterate
yourself to get to a decision framing
around where you can solve these
problems and human are irrational to a
machine. We make decisions that are
completely irrational. Every major
startup that's built or small startup is
an irrational decision. Right? I mean
you're not building something that is
certain. You're building something
highly uncertain. You're putting money
on the line. you're wasting your time
potentially, right?
>> If you're being logical, you would have
never done it.
>> Yes. If you're if you're a machine,
don't do it, right? Um because the
probability of failure is incredibly
high. And yet we do that because we're
risk engines. U you know, I mean, even
you know, I would argue every attempt at
a relationship, we're taking a risk even
though internally we're able to
calculate that. A machine would
probabilistically say, "No, don't do
that." Right?
>> That's where that's where reasoning and
critical thinking come in, right? And so
you need to realize when to use it and
when not to use it. And it does it does.
it still is based on at least today um
you know prompt and output. there isn't
an evaluation and a learning feedback
loop yet. And that's the differentiation
I think from where we're at today versus
>> I mean I think we both lived in a time
when the calculator was not allowed in
the classroom. That's right.
>> Right. And the criticism was it will
make you dumb.
>> Yeah.
>> You will know now how to do math.
>> That's right.
>> The calculator has made math not just
more abundant.
>> Exactly.
>> But the you know faster, bigger, faster
breakthroughs in math
>> because now more people can do math
>> because of the calculator and then the
computer. Right. I think AOL would do
the same.
>> I 100% agree. Yeah. I mean, it's like
transportation. I mean, you know, did it
but could you should you still keep
walking around or or take a train? You
know, like Yeah. I mean,
>> you could use a horse.
>> You can use a horse. Yeah. Exactly.
>> The last question before we wrap up is
if I want to build something,
do I increase my chance to be backed by
Animoka if I use AWS?
>> What a setup question. Um, well, okay.
So first of all, um I'm afraid to say
the answer is no.
>> Yeah, I know the answer to that already.
It was we wouldn't we wouldn't want to
know whether you're using AWS. We would
want to know how you're using a right.
So if you come to us and say I'm using
AWS with this whisp this this optimized
whatever I know the price structure I
can optimize it and blah blah blah.
>> That says something about you. Okay.
>> Right. That that's more important than
the particular tool you're using.
>> Yes. Of course.
>> Yeah. Yeah, and I I would just add to
that um you know from a different lens
uh to Yat's exact point we see a lot of
founders regardless of cloud provider um
you know making poor scaling decisions
right and so the last thing regardless
of the cloud provider that you choose as
a founder you to be going through when
you're scaling is a migration
>> and so we spent a lot of time um you
know leveraging and working with tools
that are they're using offtheshelf
systems um that that work quite well uh
but they lack the infrastructure to
provide scale and security. So um again
the last thing you want to be going
through is is a migration when you're
when you're scaling and so you need to
start thinking about regardless of the
cloud provider um you know what what
tools that you're using that are going
to allow you to to scale uh securely and
safely.
>> Hey just one quick thing in closing if
we have time. So um AWS and Anamoka are
are partnering on an Aentic uh World Cup
of football. Oh, that okay talking about
sneak peek uh stealth mode. Um but
basically uh developers um of all uh
sizes are going to be able to go through
uh a workshop to learn about how to
develop agents.
>> Um and then one of the outcomes of that
is uh a game interface that Animoka has
designed. So the agents will be
competing uh in a soccer tournament uh
in real time and we're going to be
running that at scale globally to
correspond with FIFA this year. So be on
the lookout for that. And some people
might do some polymarket bets on the
side. I think I'm sure
>> I would not be surprised. We got to
figure out a way to tokenize this.
>> Yes. Exactly. Exactly. Yeah.
>> See, I love this conversation cuz I've
learned so much from you guys. It's
always great talking to you and my god,
I've learned so much from you, Ian.
Thank you guys for, you know,
>> thank you for having us
>> being on the show and please come back
soon. I would love to have you back. You
know, there's so much more we can talk