August 12, 2025 • 50 mins
How Decentralization Is Reshaping the World
In this episode of High Stakes, host Alex Nwaka sits down with Yinal Ozkan, Principal Solutions Architect at Amazon Web Services, to unpack how decentralized technologies, blockchain, and AI are changing the institutional landscape. From AWS’s early bets on cloud and AI to today’s breakthroughs in Web3, Yinal shares firsthand insight into adoption curves, enterprise readiness, and the future of autonomous agents.
KEY TAKEAWAYS
• Early cloud skepticism mirrors today’s AI adoption hurdles
• AWS growth from $1B to $100B reflects deep infrastructure scaling lessons
• Startups excel through pragmatism and clear, short-term goals
• Web3 realization phase is driving rapid decentralized application adoption
• Public chains dominate institutional Web3 activity over private chains
• AWS balances emerging tech with core infrastructure via customer-led innovation
• Trading and exchange activity are fueling wider Web3 ecosystem growth
• Security, scalability, and integrated data management are core to AWS AI offerings
• AI adoption journeys typically move from off-the-shelf tools to fine-tuned models
• Web3 and AI converge in decentralized autonomous agents and on-chain transactions
BEST MOMENTS
00:02:21. “I was building data centers, infrastructure, main security consulting, and one of my friends said, have you ever heard of AWS?”
00:04:25. “When I started building online banks in 1995, everyone told me this will never happen.”
00:05:59. “Same thing with AI, nobody assumed it was happening until now.”
00:08:16. “In my past life, I built low latency, high frequency trading infrastructure, a lot of it on premise.”
00:12:20. “Now we are in the actual realization phase where decentralized applications are changing the world.”
00:17:18. “We came up with a service and solution set providing the most comprehensive infrastructure for our customers.”
00:20:32. “Today, retail users can directly interact with the largest centralized and decentralized exchanges, 24/7.”
00:48:45. “My agent says I can spend $0.50 today, instant transaction, and the remote agent tells my agent it’s sunny.”
🎧 New episodes every Tuesday. Tune in weekly for sharp, no-hype conversations on blockchain, AI, and enterprise innovation.
📢 Follow Validation Cloud for more insights:
• Website: validationcloud.io
• LinkedIn: linkedin.com/company/validationcloud
• X (Twitter): x.com/HighStakesWeb3
• YouTube: youtube.com/@ValidationCloud
👤 Hosted by Alex Nwaka, CSO at Validation Cloud.
Cut through the noise. Stay ahead of the market.
This podcast has been brought to you by APodcastGeek. https://www.apodcastgeek.com
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
We are in the actual realizationphase
where we can see and understandwhy decentralized transactions
and the applicationsare changing the world.
I can see actual transferof ownership or tokenization
(00:20):
or actual trading is happeningat a very different scale.
The world is changing.
It's changing very rapidly.
I'm seeing superexciting solutions every day
and I'm very happy to be here.
What's your boldest predictionfor blockchain, Web3
or AI for the rest of this year?
(00:41):
Welcome to High Stakes,the podcast
committed to demystifyingthe institutional adoption
of blockchain technology.
I'm your host, Alex Neil Walker,and today we're elevating
the Web3 discourse in a seriesof conversations with industry
iconoclast that are shapingthe future of Web3.
Now let's get started.
(01:06):
Welcome back to anotherepisode of High Stakes.
I'm your host, Alex Walker,and today I am joined by,
you know, O'Connell, who is a,principal solutions architect
at Amazon Web Services.
Welcome. You know.
Oh, thank you for inviting me.
Awesome.
So, you know,I always like to start with,
you know,sort of a very high level
question that, you know,sort of gives the audience
something to think about.
(01:28):
You joined AWS back in 2012,as the global head of financial
services technology.
And during that time,you know, AWS
was generating roughly about $1billion in annual revenue.
And we fast forward to today,it's surpassed 100 billion,
which is 100 x growth story.
That's a pretty rare journey.
(01:48):
So can you share a little bitabout both?
You know, how you and sort ofAWS have evolved over that time?
Yeah, it's is it's a journey.
You know,I remember the whole field team
for the withAWS in a hotel room discussing
what we're going to do next.
And if you know where we arein terms of field teams
and the whole AWS,that can tell you one story
(02:08):
inside that one. Yes.
I joined in 2012 on the idea,because, back
then I was a consultantin North East
for the previous ten, 12 years.
I was building datacenters, infrastructure,
main the security consulting,and I was building a data center
back then.
(02:29):
And one of my friends said,have you ever heard of AWS?
I heard them like, they're justkids play, you know?
And then I just loggedin, just looked left and right.
And yes, there were likethey were in there as today.
But the idea was great.
You know, the company whatit presented to opportunity.
So I joined in 2012without ever thinking.
(02:51):
I said, like, I just made a callthe next day you call me back.
And suddenly in 2012,I was one of the first people.
The title was architect,you know, and, they said, like,
you are now our security person.
I said, no,I'm not your security person.
This is financial services.
This.
And, so that took me like,this was a great journey
when I joinedand looking at back, you know,
(03:13):
we had the great idea.
We always said, you know,the this business
has the potential.
And with it still does.
And, what was the idea in2012 is taking
up is now the reality.
It's, almost $30 million.
Just that was the last quarterand like one under $12 billion,
(03:35):
you know, run rate.
It's a very different operationtoday.
And I witnessed wholefrom 2012 to today.
So and sort ofdrawing on that experience,
do you see any parallelsfrom the early days of cloud
adoption, to what's happeningtoday with AI adoption
within sort of the enterprise,vertical specifically?
I do, and it's like,I think that's pretty
(03:55):
much my story.
When I, I'm one of the guyswho started very early
on this whole business, notjust the cloud like, since 1995.
I'm a professional consultanton the fields, you know,
and when I startedjust extensively, like,
(04:15):
my business was building onlinebanks, you know,
and, like internet bankingand, and everybody told me,
you know, you're a nice person.
We like to talk to you,but this is not possible.
This will never, ever happen.
This is a kids storythat will be, you know, like
a few people will play with it,but that will not happen.
Fast forward around 2000, I wasin the virtualization business.
(04:39):
We were virtualizing servers.
I heard the same story, said,you know, this is really nice.
Some people can do itin the labs or left and right
the virtualization.
But this is not real.
This is not secure.
This will never, ever happen.
The same peopleand you know the story.
And same thingwith the cloud computing.
Answering your main question,when I joined all my friends,
(05:02):
all my professional peerstold me, like,
you're leaving your current joband joining Amazon
and for cloud computing,are you sure you know?
Because yes,maybe Netflix is doing it
because they are like a startupor things, but like the the
your real customers will never,ever do that.
You know?
(05:23):
And for the record, I backthen I was doing financial
services, health care, you know,very regulated industries.
And that wasn't even likecompliance and regulation didn't
allow to use cloud computing.
I know global banks just didn'teven use the word cloud.
That was like using the word inpresentations was not allowed.
And everybody said like,this will never, ever happen.
(05:44):
You know, look where it took us.
Same story.
This I don't know what it islike.
People want to don't wantto believe in the good stories.
You know,they always want to find
there must be another back storythat this shouldn't be real.
Same thing with theI also took place,
you know, and it wasn't.
(06:06):
We didn't have a lot ofbelievers.
Amazon is an early player in theI have not the gen.
I use the relativity ofwhen many people mix
those two terminologiesand when the AI came in,
the like,Amazon was like already
we were already likethe predictive analytics,
like routeoptimization recommendation
engines, you know, robotics.
(06:27):
This was already an Amazon play.
Like AWS,teams were already very familiar
with theAI technologies and things.
But when it comes to generativeAI or actually
seeing the results, the businessresults, actual results,
I had the same reaction.
Everyone said, oh, this is nice.
This is likelike some developers
or some scientistsare doing really good things,
but this will not happen.
(06:48):
The compute the data.
It's not because, you know,like technically it was proven
way, way before that,all the technologies.
But the thing is,like nobody assumed
it is happening until now.
And it is now.
The history is repeating itself.
Now we are seeing actualresults.
People are buildingapplications, actual solutions,
(07:09):
addressing real businessproblems is happening.
I'm having a some sort of wherethe solar industry is going.
Excellent.
And you know,we obviously know you
through your your work in Web3.
But as you sort of mentionedthroughout this conversation,
you had experience acrossa very broad, sort of tech
spectrum,especially financial services.
Can you give the audiencea little bit more of a sense of,
(07:31):
you know, what your role at AWSreally encapsulates today?
Maybe talk aboutsome of the work you've done,
in sort ofthe financial services vertical,
whether that's, you know, highfrequency trading, low
latency, that sort of thing,just sort of curious.
And I want the audienceto get a sense. Yeah.
I'm one of those Web3 peoplewho come from trad fi,
you know, like,that's why we call it
(07:51):
like, traditional finance now.
And, so back in my consultingcare career and first
five years at AWS, I strictlywork with, very large,
you know, financial serviceinstitutions, mainly with,
investment banks, hedgefunds and global banks,
you know, and key,like to critical
(08:13):
infrastructure companiesin financial services.
And, inmy past life, I built, like,
low latency, high frequencytrading infrastructure,
a lot of them on premise.
And that was me.
And fast forwards in 2024,you know,
AWS decided to, you know,bring more,
(08:37):
muscle power to the vertical,you know,
as a principal solutionsarchitect
with the startup segment,you know, we we had a focus on
very key startups in the space.
And I became the principalsolutions architect
for some named accounts.
What a solutionsarchitect does is
(08:57):
we are the technicalend of the conversation.
And yeah, I mean, you may think,oh, customers
when they engage with AWS preor post
sales, like they have questionsand they have a person
as an extension of their teamto answer key questions.
It's not just limited with that.
In, in the proper definition.
Well, what I do andpeople like me, what we do is
(09:20):
we work with our customersand we define the to be
architecture, the infrastructurefor a future state
which we learnfrom the best practices by our
experience, by the company,by the technology that we know.
We paintthe picture of a future state,
and every company is atanother state.
I call it the as is state.
(09:42):
And the solutions architect jobis to take company from the
as is to to be state.
And that is not a single path,a single activity stream.
There are hundredsof activities.
Some of them are likejust turning a switch.
Some of themtake like five years to migrate
to your core platform.
And what we do is I usually saywe are not more intelligent
(10:02):
than our customers,but we have more customers,
larger ears, and we learn a lot.
So we help our customers from asis to, to be architectures
and the most efficientand the fastest way.
So that ispretty much what I do.
I usually introduce myself.
I'm the technicalend of the conversation. Got it.
And you know, obviously workingwith so many different
(10:23):
types of startups,you get what I would,
you know, describe as a lensinto where the future is going.
And I think you sort of describethis in sort of the, you know,
you saw cloudthat it was going to be
a big thing,you know, before others
like really latched on to it.
And you saw sort of the marketreally expand.
Are there any startup trendsyou're seeing that enterprise
(10:44):
leadersshould be paying attention
to right now, that they are not?
Well,it has always been the case.
Enterprise leaders,including myself.
That's how I've noticed startupslike they know
startups are innovating.
You know,it is a very different mindset,
first of all, to startups.
And you, you know betterbecause you are also part of it.
You have a timeline,you have a runway,
(11:07):
you have been investedand you have a goal.
It is not likeyou're investigating
or it's not like,like, like day to day
or even a long term strategy.
You have a very well-definedpath to success.
And what I really like instartups is it's very pragmatic,
like like you dothe things that can be done
(11:27):
in the most efficient,the fastest way.
That is possible,because there is a very clear
goal in like a yearor two years, three years,
this is where I want to be.
And and you don't needto follow any infrastructure
or this is a databaseI need to migrate to.
You are starting from scratch.
There'sa whole different mindset.
So at a high level businesslike startups always come up
(11:50):
with likethe next gen of the story, okay.
And that is very natural.
That's that's how the marketswork, right?
In termsof what I have been observing
is that, you know,there are a lot of curves
that showslike the technologies,
like they are like initialand like mid stage and tags.
I mean, I thinkwith the Web3 technologies,
(12:12):
there was an initial phasewhere everyone went crazy,
there was a peakand everybody went like,
oh, okay, let's do it.
And then good or bad,like out of reputation, like,
I don't need to repeatthe stories
that everybody knows.
But now we are in the actualrealization phase
where we can see and understandwhy decentralized transactions
(12:38):
and the applicationsare changing the world.
You know, and I see a lot oflike being where I am right now.
I can seeactual transfer of ownership
or tokenization,you know, or actual trading,
you know, is happeningat a very different scale.
(12:58):
It is not very publicly,you know, advertised right now,
but if you trackjust the keywords, like it's
not just called crypto or Web3,that's the only difference
I'm observing.
As long as you followthe activities
under digital assets, you know,the world is changing.
It's changing very rapidly.
I'm seeing super excitingsolutions, every day.
(13:21):
And I'm very happyto be here, by the way.
And you know, maybe to flipthat last question on its head
a little bitfrom from the inside,
how does AWS balance exploringemerging technologies alongside
of core infrastructure?
So giving yousort of the relevant
example here in this room,you know, when blockchain
first gained traction,how did AWS decide
(13:42):
whether and how to engage?
Just from your experience?
Like first of all, the companyI really like
is governed by principlesbecause you cannot manage to
so many people.
These are the three thingswe will do.
You know,you actually put very high level
governing principles.
We call them internallythe leadership principles.
You know,the customer obsession,
(14:02):
like we have 16 of them,for the record.
And but the ones I really likefeel myself
moving more closer to our like,the customer obsession
deliver results, you know, and,so over 90% of AWS
products were built.
Not because we had an idea,you know,
(14:25):
it is actually just a reverse.
Our customers had an idea.
And AWS is very receptiveof customer feedback.
We have a lot of enginesto listen to our customer base
and take their input and buildtogether with them.
And the blockchainis a very good example.
I mean,our natural way of thinking is,
(14:47):
oh, AWS have some peoplelike some leadership,
some people technicaland some people business.
They all come together.
Okay, what are we going to doabout this blockchain?
You know, let's do something.
It is actually very different.
The decision making at AWSand we
I mean, you probably heardwe have a written culture,
you know, and the ideas actuallyfirst are they are shaped by,
(15:10):
written papers.
You know, and it's not just oneperson writing a paper on this.
Oh, that's it, let's go for it.
Even today, I'm reading like 1or 2 papers every month.
And these are like,super detailed, like papers.
And the blockchain, the decisionmaking actually happen
the same way.
(15:30):
We have so many different teams.
We have product teams,we have partner teams,
we have field teams,we have support teams.
And it's a large companywith very different leadership,
very different goals. And,but eventually we are meeting
under the same, umbrella.
So what happens is all theseideas form
into a written format.
(15:52):
Okay.
And when they comeinto a written format,
they been read by allthe peers, leadership, everybody
reads, and the good ideasjust flowed up.
And the thing is, many peoplesaying like, we act like
we act quickly as we can,a company at our size.
You know,I am also learning by time
as the companies grow, it'sa little bit different action.
(16:12):
But at the core of the thingsthe Web3, decision making
is that, like, first of all,our customers said like we
we in we're doing this,you know, and, so we said they
and they ask us exactly like,hey, AWS, you're our partner.
What are you doing?
You know, our decisionmaking process is we evaluated
all the papers, you know,and there was like,
(16:34):
a single idea.
Should we build a service?
We should serve our own?
Or should we buildan infrastructure?
Should people leverage?
Should we just stay independentout of it?
You know, should we just gointo the application layer
and just go in the paymentsbusiness, you know,
all those ideas,all in the trading compliance.
There are so many differentangles.
(16:55):
And they were they were paperswritten
in every, every other way.
And sometimeswe accepted not just one.
For example, we had Amazonmanaged blockchain as a service.
On the other hand,we build a field team
that are helping customersto build their own applications.
Or we have security teams,identity teams modify
modifying key management serviceso that like it's support, works
with the enclavesso that, you know,
(17:16):
our customers can build it up.
So there wereso many different activities.
And then the way it comesto realization
is that when those papersget into the plans, we call them
ops operating plans.
And it'sa very structured process.
We actually know what we'regoing to do next year, you know,
(17:36):
and next five years.
Like those are all differenttypes of planning
that happen at AWS.
So the first signs I rememberI is like 2017, 18, this started
to to flourish there.
Those ideas,the first papers were read.
But every year we added,every year we added different,
layers, to this mix them.
(17:58):
We try the idea,sometimes we fail, sometimes
we try certain things we can.
Oh, let's write a builda service, you know, doing like
fair market access, you know,like or like let's build like,
another exchange, like all theseideas come left and right.
And, long story short,we came up
(18:18):
with a service, solution setwhere our customers like, today.
So which is like we areproviding the most comprehensive
and broadly,you know, utilize infrastructure
for our customers.
And, you.
Know, when you, look at the Web3sort of ecosystem,
as a standalone,what would you say
are the fastestgrowing subsegments within Web3
(18:40):
or just company typesdriving significant adoption on
AWS today? In that category?
I think I really likedwhat you said
about your vantage pointbecause like, it's
very relative,because I work with a certain
set of customers, you know,and these are like in my way,
like, like growing really welland like, like and between
all types of customers,like which ones
(19:00):
I see a lot of activity in,you know, exchange and trading,
you know, and it's,it's very interesting,
like our initial conversation,the move from
traditional finance to, to DeFi,like the this,
this transition is happening,but it is not like
entirely different.
Everything is new.
Everything is different.
(19:22):
It is also following thefootsteps of the proven steps.
Like,I remember one of my customers
said, look,I came with this great idea.
Like where we do the trades,we should do trade compliance,
you know, screening so that likewe don't spend so much time.
I said likeso that is I mean that's
a great idea first of all.
But you didn't invent this,you know.
So I see huge activity and,you know, you can cross-check
(19:44):
what I'm saying with numbersis in, in trading
and this is not like,the jargon in our world is like
the retailpeople are directly interacting
with exchanges.
This is like, this has neverbeen seen for the records
like if you look at the historyof trading, like,
usually it is institutional.
(20:05):
Even if the individualstrade like us,
they are usually tradingthrough brokers.
Some intermediate, companies,and it only happens in certain
hours of the day, you know, incertain places, certain things.
Now fast forward today it is724 retail.
People can literally interactwith largest centralized
(20:26):
and decentralizedexchanges directly.
And also the institutionalplayers are not coming in
because there's a liquidity.
Is there likepeople are interacting.
I'm super excited to and,it may be impacted where I am
because I'm building a lot of,low latency systems, you know,
but it is happening right nowall over the world,
certain parts of the globemore than the others.
(20:48):
And we all know like it'sa little bit eastern way, but,
hopefully in U.S soon and,but the trading is happening.
I think it will fuelother things because once it is,
financially stable world,I expect more, data providers,
I expect more DApps, I expectmore chains doing like,
(21:08):
like day to day, things as well.
And this is separatedfrom what's going on
with the stablecoinsand the tokenization that's also
going where super active.
But I think, in my like, ifyou relook at the growth,
charts, I think I really likethe trading part.
(21:29):
And then, you know,there's this constant debate
in sort of the what'scalled the institutional realm
and their interest in Web3,between different types
of implementations.
Whether that is, you know,blockchain networks or private,
you know,permissioned implementations.
I'm curioushow you would frame that,
you know, sort of thinkingfrom an institutional
perspective and,you know, maybe pros and cons
potentially just sort ofwhere do you see it
(21:51):
playing out over time?
It's interesting.
Again, if we do a correlationwith the past terminology.
Rememberthe first days of the cloud?
There was a big conversationabout public cloud.
Private cloud.
Or I want private,I want public and fast forward.
Like it's so like my my first,you know, interaction with the
the Web3 world or public chains,private chains, enterprise
(22:13):
party, you know, retail likelike all those conversations.
But I think the story willis going at the same direction.
And all my activities, even if,like, institutional,
like large scale enterprise,is currently warming up.
That's the best way to put it.
With the with the vertical,it is all, happening
(22:36):
over, public chains.
There is definitely activityin the private chains,
if you ask me.
I'm in my personal view,like 90, 95,
maybe even more of the activityI'm seeing is in the, in the,
in the publicside of the business.
They are still institutional.
They are still large players,traditional, financial players.
(22:58):
But the transactions are takingplace on the public chains.
Got it.
Now, obviously,like our space, Web3
is really defined by community.
And so, beyond sort of tech,can you describe a little bit
about whatthings AWS does to support
Web3 companies from a businessand an ecosystem perspective
(23:19):
and helping them, you know,sort of collaborate and grow
within, within their own bounds?
It's a good question becauseI meet with a lot of customers,
like companies like that'susually the first question,
how can you help?
Like, are you here toto position on AWS services?
Like, do you want to show mewhat kind of products you have?
You know, that's usually a goodstart of the conversation.
(23:41):
But our answeris very different.
Like we are a part of your team,we are here to grow
your business because it'sa very natural business.
If we are partners in this one,you know, then the growth story,
we we grow together.
It's not like we just AWSgrows more, you grow less.
Technically, you should begrowing way faster than AWS.
So we are actually growingvery healthy.
(24:02):
You know, where you health away.
So in orderto accomplish the growth,
you know we work togetherand it's not just us providing
infrastructure servicesto make you more cost efficient.
And reliable or secure.
You know, it is also aligningyour business goals
and realize them together.
And this involveswith a lot of programs.
(24:23):
We have different programswith different, for companies
at different stagesof their journey.
I mean the startups.
So with the startups,if the moment they get into
the startup ecosystem as a partof an accelerator incubator
or one of the investor,I call to founded or funded,
you know, we directly startinvesting in the startups
(24:47):
and everybody is familiarwith the credits.
We give them, likeliterally tangible, support.
But it is just, like the visiblepart of the iceberg.
We support them with every,every definitive,
like way of like,like giving them
technical resources,giving them blueprints and,
(25:08):
doing joint GTM activitiesand creating opportunities
for themto reach their main audience
using our connectivityor like, like our resources.
We are literallyhaving a full motion
of growing our startupscustomers.
And, this is just oneside of the business.
We also follow the regularAmazon flywheel.
(25:28):
Like,you know, we have a marketplace
or we have a partnershipprogram.
We have the most vibrant partnerprogram, 140,000 partners
in the ecosystem.
It's it's a community itself.
And so,we, we bring in our customers
into this, this ecosystem.
So, so, so, so we grow together,like I heard,
(25:50):
I need to cross check that fact.
Like 40% of the AWSrevenue comes from companies.
We were once a startup, so.
So we are investingin the startups with programs,
with teams, with resourcesat every very different level.
And this is repeating itselfat many different stages
at the enterprise level, fortune500 level, the global level,
(26:10):
we are building dedicated teams,to support our customers.
Very helpful.
And, maybe moving intowhat AWS is doing
and has built in sort of theAI specific, you know, vertical.
You obviously have a very broadand growing, AI portfolio.
Whether that's, you know,SageMaker, bedrock.
(26:32):
Can you walk the audiencethrough how these offerings
align with different types ofcompanies building and AI today?
Yeah,very good question. Because,
we wantto have the most comprehensive,
solutionset where customers can build
and use the technologies.
As I said, theAI technologies have been with
(26:52):
AWS for a very long time.
But if we do a likea like, truly
the main focus is on the AI.
You know, I will startwith that part, you know, so,
we first of all, we providethe most performant and cost
efficient infrastructurefor our customers to build,
you know, to build foundationmodels, deploy them, train them,
(27:16):
and, run their operations.
This is like the compute,the storage, the the well-known
components of AWS is out there,and it's with the, four,
four main layers.
So first of all,we give our customers choice,
you know, a lot of differentchoices.
Nvidia, Nvidiait is AMD, AMD it is
and we have our own processes.
(27:39):
We have our own silicon,we have our own GPUs.
We give very different optionsfor our customers
so that, you know,they can build by their choice,
not because we force themor like they are forced
to do certain decisions.
At thesecond layer, what we do is
as you get into any AI project,the technology is great,
it is creative.
(28:01):
It is like like groundbreaking,but without data
it doesn't work.
So you need to have a very solidplatform to manage your data.
So, what we are doing very hardis that we are also
integrating the data layerwith the AI projects.
(28:22):
And I will explain a bitmore like
like how we differentiatethose, those parts.
But at a very high level,we make sure the data
management, the data lakes,accessing
data storage of the data,processing, labeling,
all those parts are actuallywithin the same ecosystem.
They are all integratedinto your projects.
That's the data part.
And two more pieces actuallycomplement the AWS, logic.
(28:47):
And maybe startshould be everywhere.
But again, and securitybecause security is somehow
when you are super busy,especially with startups
like superexcited about solution like,
oh, let's do itand everybody's on the on board.
But security is hard.
It is.
It requires very specific,very dedicated resources.
So we dedicate, we put a lot ofsecurity resources
(29:09):
and we make sure the securityis managed,
at least with the withthe resource that we provide,
we make it easier.
I'm not saying easy, but easier,better and more secure.
Of course.
And and the last thingthat is the bread and butter of
AWS is scaling.
So how we can scale.
So going back to youroriginal question sorry for
(29:30):
but this is, this is the, the,the managing, logic.
So, what we dois, as I mentioned,
we provide the,the most cost efficient
and performancedriven infrastructure.
Okay.
So this includes the compute.
And we alsoif you want to have a
to have thisinfrastructure managed,
you know I'm buildinga foundational model.
(29:50):
I'm being likelarge inference, farm.
So all those thingswe have a managed product
called SageMaker.
And SageMaker isa family of products actually,
with the SageMaker,we just really is a unified, ID
so, do unified studioactually allows you to access
all these subcomponents,the data layer and the AI layer
at the same time.
(30:13):
And it helps you to,build, train, you know, deploy,
manage, and MLOpsall in under the same umbrella
while managing your data sets.
So that's the SageMakerside of the, solutions.
So this is for companieswho are either directly
accessing,tapping the infrastructure
(30:33):
or building the infrastructureor managing all your, like,
process through a managed layer,which we call SageMaker.
All the notebooks, like all the,all the ops there, they,
they all come with the SageMakerone step further.
If you go, we have the bedrock.
If you are buildingapplications,
(30:53):
you know, you are not buildingfoundational models.
You still do like your own datasets.
You do rac you know, youyou are even like,
like fine tuning certain modelsand you need to access
different models,you know, you need to test them,
you need to optimize them.
And, what we do is,we provide access to multiple
LMS and foundational modelsthrough bedrock, through an API
(31:15):
like that is like, actually,if you build applications once
this is a really great service.
You areyou know, what you want to do.
You want to leveragethe whole technology.
You want to use the technology,you want to build new things
and we have the bedrock layerwhere you have, access
to all this infrastructurewith certain things
(31:37):
I mentioned at the beginningwith the security,
you know, and also,like we have guardrails
so you can control like,like, questions and answers.
And with the agents integration,we will probably touch that
topic later.
And all those, partscome through an API
through bedrock.
So it's easierto build applications,
(31:58):
but it's not there.
We still goin the application layer,
because what if you are notbuilding applications, but
you will still want to leverage,you know, the the technology
we havelocked so many, solutions and
like how your work gets donewith the generative AI.
So, on that layerwe have the cube business.
(32:19):
We you basicallydrop all your corpus,
the data into the solution,and you just start querying
your data sets throughnatural language.
You know, that is doable.
We have two developers.
So you want to build,you turn to your developers
using the assisteddevelopment environment CLI.
Q developer like these are greattools.
(32:40):
It's not limited with thateven from from supply chain to,
call centers.
We built in the AI technologyinto the solution set
so our customerscan use the new technology.
Easily.
So like,like from the user level
to applicationbuilding to the foundation
infrastructure layer,we have three different
sets of products.
(33:01):
You know, I thinkone of the things
that many companiesare going through
the thought process of,and I think you're very well
positioned to maybe opine onwhat are the pros and cons,
or at leasthow to how to think about it.
There are a bunch of differentmethods for building on top
of foundational models, right?
So you have prompt engineering,you have fine tuning of models,
so on and so forth.
How do you think companiesshould be thinking
(33:21):
about the best path to pursue?
As they think about buildingon top of foundational models,
is theresort of a certain framework
you like to go with,or is it much more
sort of company dependent?
One size doesn't fit all.
That's what we learned,you know, and,
it is very dependent onthe resources and the timeline
(33:41):
and the expectations.
This is notjust the, you know, speaking,
you know, and,I've seen our customers
going through this journey,you know, oh, I will increase
my, you know, productivity.
I will increase my insights.
You know, I willI will use generative AI,
but that depends onwhat kind of resources you have,
(34:02):
what kind of timeline you have.
And we learn togetherwith our customers.
You know,usually the path is like
from easier to more difficult,you know, that's that's
one way of I can put things up,like one of them is directly
usingalready built systems, like,
like we have a lot of, you know,solutions, as I mentioned,
like the connect and,you know, the Q business,
(34:23):
like ready to use next day.
You just plug in the chat, the,the the querying.
That all works.
You know, that's one areathat's easiest.
And then you startbringing your own data
sets, the serac, you know,that's usually,
you know, the next step.
You just say, oh, I like it.
But I want to do it securelywith my own data sets, you know,
(34:44):
and suddenlyyou are one step closer
and now you're interactingwith your own data.
In addition to what lumberingzero the foundation model breaks
you, you know, one step down,and you said, okay,
I like this, this lamb, but,you know, it is.
I want to actually finetune this with my own,
you know, theI want the answers this way,
(35:07):
you know, like,I want it differently.
Fine tuning comes next,you know?
So that's, that's requires.
And when you switchfrom actually rag
to fine tuning,I think there's a big step
there, you know,then you really need
to understand the technology waybetter with deeper
and with more,like more technical resources.
Again, with the tool sets,we are there to help,
(35:28):
you know, definitely tryingto make the journey easier
for our customers.
And at the end of the spectrum,we have the builders
of foundational models, whichwe work with them daily as well.
And I think in terms of numbers,I think it's like a funnel.
But, the activity is,almost everywhere,
and it's a learning process.
(35:48):
Some customers just jump inand it's the wrong place,
wrong time.
But it depends on the skill setand the timeline
of our customers.
And and what do you think aboutwhat enterprises are
are doing inAI as it relates to,
the ability to marshal resourcesboth in terms of team size
and data sets.
(36:09):
How is that shifting overthe last 1 to 2 years, meaning
we're seeing massive improvementin sort of the quality of models
and new standardsfor connecting different types
of data sets into models,such as anthropic, MTP.
So do you view it as, you know,these models are getting
so much better that it requiresless people and like sort of a
(36:33):
a smaller critical mass of datato make them useful or is
or is that an assumption?
Just based on, you know,sort of the improvement
of models,you still sort of need
a critical mass of team and datato really get to a place
where value is being created.
Yeah.
In my personal experience,I think, it is going in
two ways.
(36:53):
You are correct,it is getting easier.
The models are gettingmore advanced, you know,
so if we were doingthe same chat bot that we did
like four years ago,it is like literally
light years ahead right now.
You have the chatbot in a few hours,
you know, with your own datayou would like if you use Q
you know, but the expectationsare getting more complex now.
(37:16):
Now our customers are askingfor more now,
like we want to addlike like a very good
question is like, okay,how can I merge the smart
contracts I'm using with myAI agents?
It is a very different questionthen, you know, just,
you know, I want to buildthe chat interface.
(37:38):
You know,I think the time and the effort
is not decreasing,maybe increasing the resources,
but the deliverablesare increasing
in an exponential way.
So there is there is no like,like a linear line of okay,
it's easier.
So I need less resources.
You know, the expectationsare getting higher.
(37:58):
I think the allocationof resources are changing.
I think everybody'sobserving it.
So theallocation of resources to
projects, taggedwith AI are very different.
And this is very normalbecause there is a business
opportunity.
But, my personal opinion is thatthis is not a separate task,
(38:21):
like the security conversationwe had two years ago.
Oh, you should havea security team.
They should do theirsecurity stuff, and we do it.
The business,you know, it is not like that.
So if you are in, like I don't,I don't know,
like in the legal business,if you are practicing law,
you know you areyou have a gene I project
if you are in financial servicesyou have a gene I project.
(38:42):
It is independent of the teamsor like the the vertical.
You're in every business.
This is like integratedin the core of the business.
This is not like a separate teamof AI people.
They are doing great thingsin the lab or in the production.
It has no it is being integratedin the core technology of the
companies and technologiesis becoming a core business.
(39:04):
Like multiple, transitionsare happening at the same time.
Got it.
And you startedto go down this path.
So I think if we canwe should ask this question now,
which is, you know,there's a lot of attention
being paid to agents andautonomous, you know, workflows.
I thinkjust let's start definitionally
because I thinka lot of the audience
may not really know what a truedefinition is of an agent.
(39:26):
So can you give the audiencesort of your definition of like,
what is an agent?
And then in addition to that,maybe some impactful
examples or implementations thatyou've seen of agents today?
Okay.
So yeah, first of all,there is no true definition
because it's just evolvingevery day.
You know, at the end of the day,the agents are small
(39:50):
applicationsdoing tasks for you.
But what definesan agent versus not an agent
is a good conversation.
So the way, I look atthings is like there must be
certain characteristics,you know, in the small
application that we are usingand to be called an agent,
you know,and I think these are like
getting real settled,settling down. Okay.
(40:14):
And, the first one is definitelyneeds to be autonomous.
You know,it needs to be operating
without human supervision.
Okay.
And the second thing,and there's also like
a longer conversational agent,like I was agents because
the agents let's focus on thetopic is actually goal oriented.
(40:34):
They have a task like,you know, to X do y.
So they are goal oriented.
So let's continue.
Autonomous goal oriented.
And the third one isthey're interactive.
This is very interesting.
It is not like an API call.
Go and do this call.
No, it's not likethe agents are interactive.
(40:57):
They get the response basedon the response they answer.
So they are autonomous,goal oriented, interactive. And
the fourth one is they areactually
using their environment.
I read in some place is calledit's a prescriptive.
So they get the weather datafrom left,
they read a file from herethey go to a database
(41:17):
and query a database.
You know, they are perceivingtheir environment,
you know, and they are gettingthe data themselves.
And and theAI part is the last part,
which is they are adaptive,they are learning
and they are changing,you know, so fast forward
whateverI told you, like autonomous,
goal oriented, interactive,perceptive and adaptive.
(41:37):
You know, these are the partsof the small application
which is doing things for me.
So, in order to be called agent,the components are, you know,
first of all, it collects data.
You know, it's on its own.
Okay. It does reasoning,you know, this is key parts
like reasoning.
(41:58):
Not too many people get like,I have a task.
I been runningthis task for years,
but my tasks were doingwhat I told them.
You know, the agents, onthe other hand, are reasoning.
I got this answer.
This is what I'm doing.
And, and also not just reasoninglike my day quickly I'm using
they are taking actions.
Oh I see, and you're using allthese files in the wrong folder.
(42:20):
Can I put them into the rightfolder? Like literally.
And if I say yes, that's veryby way risky move on my laptop.
My, my Shukla is telling thisthings.
You know by the way we have theCLI tells you a lot of things.
It's a very exciting times.
But I'm also very super scared.
So like I say,I will move all your files.
I will create youa GitHub project.
I will create a readme file.
(42:41):
Are you okay with this?
It's like okay, I'm ready,but it's a very different times.
For the things again,like the perception,
the reasoning, the actions,and the last two parts
for the components,they remember things.
It has a memory, you know, andthey are adaptive and learning.
So these are the componentsof the agents.
(43:03):
Those are the small programsthat we are using.
And there is not a single daywhere we do not come up with
new ideas for the last two,three months lately.
So got it.
We'll talk a little bit aboutthe intersection of Web3 and AI.
You know, we're seeing major,major breakthroughs,
you know, globally on unsortedAI specifically
and sort of continuedexperimentation in Web3.
(43:26):
How do you see the AI revolutioninfluencing
or accelerating progressin the Web3 space?
Well, first of all,let's remove the obvious parts
where very quickly,you know, the first one
is the development,which is queue
and sometimes gets very complex.
You know,the development will be easier.
Many people skip this part.
You know,if you have never build with AI,
(43:50):
you know, you don't know the thethe impacts of like the speed
and the the options that theAI assisted code development.
And now I think it was a blockerfor the growth of the ecosystem.
What we are working inis like finding the
the skills and developersand the things that is changing.
(44:11):
Okay, that is very obviousand it's forcing everywhere
to go,but also very specific to Web3.
We are seeing this,speeding up in the development.
Okay.
The projects are comingfaster online
and with more features,more things.
Like, that is very exciting.
Second thing, the most obvioususe case is decentralize
(44:31):
autonomous agents. You know,and there are frameworks
that we can talk about.
There are protocols like,like like like, Ocean protocol,
like the singularity, you know,like there are so many things
the this is very normal.
I have an agent,and this agent is autonomous.
It's self-directed.
It is doing things for me. Okay.
(44:53):
And especiallyin the trading space.
This is this was super exciting.
You know, we have seen plentyof interesting projects there.
And that was very obvious. Okay.
But if you ask me, I mean, I cancount so many different projects
and things, but let me explainwhat's happening in the space.
Okay.
In the spaceI work with the startups.
(45:13):
These are like like we call themengaged, like like startups
really investing in this.
The first step I seeis that, look, all these labs,
when they were trained,you know, they were in trained
with the chain data.
You know, for some reason,I don't know if this is public
data technically. You know,I don't know why.
I don'tknow the reasoning behind it.
(45:34):
So, so suddenly all our peerscame up with the idea.
Look,you could do a lot of things
with the exceptionof the chain data.
So and then, you know,you know, the data providers
and everybody likeand suddenly everybody is asking
the same question, look,we need to fix this first.
We need to make sure thisnatural language interaction
and like, like,like getting all this big
(45:56):
weave of the whole data setsshould include the chain data.
So we are seeing a lotof solutions coming up online,
as like a broker advisorto investor and vice.
So likelike so many different MPs.
So like the, the,the projects that are leveraging
AI technology to work with the,the chain data.
Okay,that's one side of the projects,
(46:18):
you know,and that's super active. Super.
I'm super excited.
I think that will changehow people interact
with the chain,you know, and query the chain.
So that's one end of the things.
Personallythere is a big like I mean,
you can call me the geekor like I'm on the technical.
And so what is making yousuper excited is like
(46:39):
I work with the fintechauto financial service.
And usuallywhen we talk about Gen
they say, look, that is likeif you ask the same question
ten times, ten timesyou get a different answer.
You know, this is actuallyfinancial services.
Not a very exciting thing.
You know, it'sactually very scary.
And the non-deterministic futureof the Gen I, you know, it
didn't work out well in the corebusiness of the fintechs,
(47:01):
you know, but fast forward,what happens if you attach
a smart contractto an agent, okay.
And suddenlythe world is changing, you know,
like now we are talkingabout traceability, like
like absolute transactionson a public ledger.
You know, we are seeingvery different use cases.
(47:22):
So, I think we will seemore and more,
applications on this,a good use cases like,
like we saw with the agents.
Let's continue with the agents.
You know, like, for example,my agent is asking for weather
data, you know,to the other agent, you know, no
humans involved or autonomous,goal oriented.
(47:43):
I want to know the weather,you know?
And the thing is, like,first of all, ownership of the
the confidential dataand the transfer of the data,
that is, that can be done.
And we all knowpeople working on this word
like the, the,the transfer of the data
ownership can be donevery easily with the, you know,
with the Web3 technologiesand one step forward,
(48:07):
I think this is mypersonal expectation and it's,
I think it's happening isnobody's talking about
the financial value exchangewhen all this AI world is
developing, growing every day.
And and I think we will see moreand more projects where I see
I ask for the weather andmy agent says I can spend like
(48:30):
$0.50 today, you know, and I'mwilling to spend 0.1 cents
on the weather dataand like to zero to that.
The Coinbase camewith like like it's an API base.
I think it will soon be likeall on the chain transactions.
That's my personally superexcited, which is oh, okay.
You want to learn the weather?
Just you have to first pay me,you know, and my agent, bound
(48:53):
by the contract says, okay,I have a dollar,
I have $0.50 today.
I could give you a 0.1 cents.
Here is a transaction.
Instant transactions.
You know, almost ish.
We know how fast the chains getand suddenly the transaction
happens.
Money is exchange,and the remote agent tells
my agent it's sunny, you know,and that is the future.
(49:17):
And we see a lot of projects,like there are like, avatars
and like, tokenization and,you know, like,
there are so many different,exciting things.
I think we are, you know,I can speak for the hours,
but we will seemore. Definitely.
We will see more and moresolutions. Great.
Well, you know,we ask all of our guests,
for their high stakes hot take,which, you know, is
(49:38):
what's your boldest predictionfor blockchain, Web3 or AI?
For the rest of this year, 2025?
I mean, prediction or wish it'svery close.
You know, I work with the U.Sbased accounts, so,
I want to see more players in usand, hopefully,
it will be a better environmentfor, for the customers.
(49:59):
And we will seemore, more activity, in us.
So awesome.
Well, with that,I will park it there.
Thank you for joining us.
You know, again, you know Icon.
Thank you so much.
It take forever me absolutely.
Keep exploringand keep innovating.
Learn more at Validation CloudIO and follow us on X
(50:21):
and LinkedIn to stay connectedas we continue to shape Web3
for institutions.
This is high stakeswithout the Walker signing on.
We are in the actual realizationphase
where we can see and understandwhy decentralized transactions
and the applicationsare changing the world.
I can see actual transferof ownership or tokenization
(00:20):
or actual trading is happeningat a very different scale.
The world is changing.
It's changing very rapidly.
I'm seeing superexciting solutions every day
and I'm very happy to be here.
What's your boldest predictionfor blockchain, Web3
or AI for the rest of this year?
(00:41):
Welcome to High Stakes,the podcast
committed to demystifyingthe institutional adoption
of blockchain technology.
I'm your host, Alex Neil Walker,and today we're elevating
the Web3 discourse in a seriesof conversations with industry
iconoclast that are shapingthe future of Web3.
Now let's get started.
(01:06):
Welcome back to anotherepisode of High Stakes.
I'm your host, Alex Walker,and today I am joined by,
you know, O'Connell, who is a,principal solutions architect
at Amazon Web Services.
Welcome. You know.
Oh, thank you for inviting me.
Awesome.
So, you know,I always like to start with,
you know,sort of a very high level
question that, you know,sort of gives the audience
something to think about.
(01:28):
You joined AWS back in 2012,as the global head of financial
services technology.
And during that time,you know, AWS
was generating roughly about $1billion in annual revenue.
And we fast forward to today,it's surpassed 100 billion,
which is 100 x growth story.
That's a pretty rare journey.
(01:48):
So can you share a little bitabout both?
You know, how you and sort ofAWS have evolved over that time?
Yeah, it's is it's a journey.
You know,I remember the whole field team
for the withAWS in a hotel room discussing
what we're going to do next.
And if you know where we arein terms of field teams
and the whole AWS,that can tell you one story
(02:08):
inside that one. Yes.
I joined in 2012 on the idea,because, back
then I was a consultantin North East
for the previous ten, 12 years.
I was building datacenters, infrastructure,
main the security consulting,and I was building a data center
back then.
(02:29):
And one of my friends said,have you ever heard of AWS?
I heard them like, they're justkids play, you know?
And then I just loggedin, just looked left and right.
And yes, there were likethey were in there as today.
But the idea was great.
You know, the company whatit presented to opportunity.
So I joined in 2012without ever thinking.
(02:51):
I said, like, I just made a callthe next day you call me back.
And suddenly in 2012,I was one of the first people.
The title was architect,you know, and, they said, like,
you are now our security person.
I said, no,I'm not your security person.
This is financial services.
This.
And, so that took me like,this was a great journey
when I joinedand looking at back, you know,
(03:13):
we had the great idea.
We always said, you know,the this business
has the potential.
And with it still does.
And, what was the idea in2012 is taking
up is now the reality.
It's, almost $30 million.
Just that was the last quarterand like one under $12 billion,
(03:35):
you know, run rate.
It's a very different operationtoday.
And I witnessed wholefrom 2012 to today.
So and sort ofdrawing on that experience,
do you see any parallelsfrom the early days of cloud
adoption, to what's happeningtoday with AI adoption
within sort of the enterprise,vertical specifically?
I do, and it's like,I think that's pretty
(03:55):
much my story.
When I, I'm one of the guyswho started very early
on this whole business, notjust the cloud like, since 1995.
I'm a professional consultanton the fields, you know,
and when I startedjust extensively, like,
(04:15):
my business was building onlinebanks, you know,
and, like internet bankingand, and everybody told me,
you know, you're a nice person.
We like to talk to you,but this is not possible.
This will never, ever happen.
This is a kids storythat will be, you know, like
a few people will play with it,but that will not happen.
Fast forward around 2000, I wasin the virtualization business.
(04:39):
We were virtualizing servers.
I heard the same story, said,you know, this is really nice.
Some people can do itin the labs or left and right
the virtualization.
But this is not real.
This is not secure.
This will never, ever happen.
The same peopleand you know the story.
And same thingwith the cloud computing.
Answering your main question,when I joined all my friends,
(05:02):
all my professional peerstold me, like,
you're leaving your current joband joining Amazon
and for cloud computing,are you sure you know?
Because yes,maybe Netflix is doing it
because they are like a startupor things, but like the the
your real customers will never,ever do that.
You know?
(05:23):
And for the record, I backthen I was doing financial
services, health care, you know,very regulated industries.
And that wasn't even likecompliance and regulation didn't
allow to use cloud computing.
I know global banks just didn'teven use the word cloud.
That was like using the word inpresentations was not allowed.
And everybody said like,this will never, ever happen.
(05:44):
You know, look where it took us.
Same story.
This I don't know what it islike.
People want to don't wantto believe in the good stories.
You know,they always want to find
there must be another back storythat this shouldn't be real.
Same thing with theI also took place,
you know, and it wasn't.
(06:06):
We didn't have a lot ofbelievers.
Amazon is an early player in theI have not the gen.
I use the relativity ofwhen many people mix
those two terminologiesand when the AI came in,
the like,Amazon was like already
we were already likethe predictive analytics,
like routeoptimization recommendation
engines, you know, robotics.
(06:27):
This was already an Amazon play.
Like AWS,teams were already very familiar
with theAI technologies and things.
But when it comes to generativeAI or actually
seeing the results, the businessresults, actual results,
I had the same reaction.
Everyone said, oh, this is nice.
This is likelike some developers
or some scientistsare doing really good things,
but this will not happen.
(06:48):
The compute the data.
It's not because, you know,like technically it was proven
way, way before that,all the technologies.
But the thing is,like nobody assumed
it is happening until now.
And it is now.
The history is repeating itself.
Now we are seeing actualresults.
People are buildingapplications, actual solutions,
(07:09):
addressing real businessproblems is happening.
I'm having a some sort of wherethe solar industry is going.
Excellent.
And you know,we obviously know you
through your your work in Web3.
But as you sort of mentionedthroughout this conversation,
you had experience acrossa very broad, sort of tech
spectrum,especially financial services.
Can you give the audiencea little bit more of a sense of,
(07:31):
you know, what your role at AWSreally encapsulates today?
Maybe talk aboutsome of the work you've done,
in sort ofthe financial services vertical,
whether that's, you know, highfrequency trading, low
latency, that sort of thing,just sort of curious.
And I want the audienceto get a sense. Yeah.
I'm one of those Web3 peoplewho come from trad fi,
you know, like,that's why we call it
(07:51):
like, traditional finance now.
And, so back in my consultingcare career and first
five years at AWS, I strictlywork with, very large,
you know, financial serviceinstitutions, mainly with,
investment banks, hedgefunds and global banks,
you know, and key,like to critical
(08:13):
infrastructure companiesin financial services.
And, inmy past life, I built, like,
low latency, high frequencytrading infrastructure,
a lot of them on premise.
And that was me.
And fast forwards in 2024,you know,
AWS decided to, you know,bring more,
(08:37):
muscle power to the vertical,you know,
as a principal solutionsarchitect
with the startup segment,you know, we we had a focus on
very key startups in the space.
And I became the principalsolutions architect
for some named accounts.
What a solutionsarchitect does is
(08:57):
we are the technicalend of the conversation.
And yeah, I mean, you may think,oh, customers
when they engage with AWS preor post
sales, like they have questionsand they have a person
as an extension of their teamto answer key questions.
It's not just limited with that.
In, in the proper definition.
Well, what I do andpeople like me, what we do is
(09:20):
we work with our customersand we define the to be
architecture, the infrastructurefor a future state
which we learnfrom the best practices by our
experience, by the company,by the technology that we know.
We paintthe picture of a future state,
and every company is atanother state.
I call it the as is state.
(09:42):
And the solutions architect jobis to take company from the
as is to to be state.
And that is not a single path,a single activity stream.
There are hundredsof activities.
Some of them are likejust turning a switch.
Some of themtake like five years to migrate
to your core platform.
And what we do is I usually saywe are not more intelligent
(10:02):
than our customers,but we have more customers,
larger ears, and we learn a lot.
So we help our customers from asis to, to be architectures
and the most efficientand the fastest way.
So that ispretty much what I do.
I usually introduce myself.
I'm the technicalend of the conversation. Got it.
And you know, obviously workingwith so many different
(10:23):
types of startups,you get what I would,
you know, describe as a lensinto where the future is going.
And I think you sort of describethis in sort of the, you know,
you saw cloudthat it was going to be
a big thing,you know, before others
like really latched on to it.
And you saw sort of the marketreally expand.
Are there any startup trendsyou're seeing that enterprise
(10:44):
leadersshould be paying attention
to right now, that they are not?
Well,it has always been the case.
Enterprise leaders,including myself.
That's how I've noticed startupslike they know
startups are innovating.
You know,it is a very different mindset,
first of all, to startups.
And you, you know betterbecause you are also part of it.
You have a timeline,you have a runway,
(11:07):
you have been investedand you have a goal.
It is not likeyou're investigating
or it's not like,like, like day to day
or even a long term strategy.
You have a very well-definedpath to success.
And what I really like instartups is it's very pragmatic,
like like you dothe things that can be done
(11:27):
in the most efficient,the fastest way.
That is possible,because there is a very clear
goal in like a yearor two years, three years,
this is where I want to be.
And and you don't needto follow any infrastructure
or this is a databaseI need to migrate to.
You are starting from scratch.
There'sa whole different mindset.
So at a high level businesslike startups always come up
(11:50):
with likethe next gen of the story, okay.
And that is very natural.
That's that's how the marketswork, right?
In termsof what I have been observing
is that, you know,there are a lot of curves
that showslike the technologies,
like they are like initialand like mid stage and tags.
I mean, I thinkwith the Web3 technologies,
(12:12):
there was an initial phasewhere everyone went crazy,
there was a peakand everybody went like,
oh, okay, let's do it.
And then good or bad,like out of reputation, like,
I don't need to repeatthe stories
that everybody knows.
But now we are in the actualrealization phase
where we can see and understandwhy decentralized transactions
(12:38):
and the applicationsare changing the world.
You know, and I see a lot oflike being where I am right now.
I can seeactual transfer of ownership
or tokenization,you know, or actual trading,
you know, is happeningat a very different scale.
(12:58):
It is not very publicly,you know, advertised right now,
but if you trackjust the keywords, like it's
not just called crypto or Web3,that's the only difference
I'm observing.
As long as you followthe activities
under digital assets, you know,the world is changing.
It's changing very rapidly.
I'm seeing super excitingsolutions, every day.
(13:21):
And I'm very happyto be here, by the way.
And you know, maybe to flipthat last question on its head
a little bitfrom from the inside,
how does AWS balance exploringemerging technologies alongside
of core infrastructure?
So giving yousort of the relevant
example here in this room,you know, when blockchain
first gained traction,how did AWS decide
(13:42):
whether and how to engage?
Just from your experience?
Like first of all, the companyI really like
is governed by principlesbecause you cannot manage to
so many people.
These are the three thingswe will do.
You know,you actually put very high level
governing principles.
We call them internallythe leadership principles.
You know,the customer obsession,
(14:02):
like we have 16 of them,for the record.
And but the ones I really likefeel myself
moving more closer to our like,the customer obsession
deliver results, you know, and,so over 90% of AWS
products were built.
Not because we had an idea,you know,
(14:25):
it is actually just a reverse.
Our customers had an idea.
And AWS is very receptiveof customer feedback.
We have a lot of enginesto listen to our customer base
and take their input and buildtogether with them.
And the blockchainis a very good example.
I mean,our natural way of thinking is,
(14:47):
oh, AWS have some peoplelike some leadership,
some people technicaland some people business.
They all come together.
Okay, what are we going to doabout this blockchain?
You know, let's do something.
It is actually very different.
The decision making at AWSand we
I mean, you probably heardwe have a written culture,
you know, and the ideas actuallyfirst are they are shaped by,
(15:10):
written papers.
You know, and it's not just oneperson writing a paper on this.
Oh, that's it, let's go for it.
Even today, I'm reading like 1or 2 papers every month.
And these are like,super detailed, like papers.
And the blockchain, the decisionmaking actually happen
the same way.
(15:30):
We have so many different teams.
We have product teams,we have partner teams,
we have field teams,we have support teams.
And it's a large companywith very different leadership,
very different goals. And,but eventually we are meeting
under the same, umbrella.
So what happens is all theseideas form
into a written format.
(15:52):
Okay.
And when they comeinto a written format,
they been read by allthe peers, leadership, everybody
reads, and the good ideasjust flowed up.
And the thing is, many peoplesaying like, we act like
we act quickly as we can,a company at our size.
You know,I am also learning by time
as the companies grow, it'sa little bit different action.
(16:12):
But at the core of the thingsthe Web3, decision making
is that, like, first of all,our customers said like we
we in we're doing this,you know, and, so we said they
and they ask us exactly like,hey, AWS, you're our partner.
What are you doing?
You know, our decisionmaking process is we evaluated
all the papers, you know,and there was like,
(16:34):
a single idea.
Should we build a service?
We should serve our own?
Or should we buildan infrastructure?
Should people leverage?
Should we just stay independentout of it?
You know, should we just gointo the application layer
and just go in the paymentsbusiness, you know,
all those ideas,all in the trading compliance.
There are so many differentangles.
(16:55):
And they were they were paperswritten
in every, every other way.
And sometimeswe accepted not just one.
For example, we had Amazonmanaged blockchain as a service.
On the other hand,we build a field team
that are helping customersto build their own applications.
Or we have security teams,identity teams modify
modifying key management serviceso that like it's support, works
with the enclavesso that, you know,
(17:16):
our customers can build it up.
So there wereso many different activities.
And then the way it comesto realization
is that when those papersget into the plans, we call them
ops operating plans.
And it'sa very structured process.
We actually know what we'regoing to do next year, you know,
(17:36):
and next five years.
Like those are all differenttypes of planning
that happen at AWS.
So the first signs I rememberI is like 2017, 18, this started
to to flourish there.
Those ideas,the first papers were read.
But every year we added,every year we added different,
layers, to this mix them.
(17:58):
We try the idea,sometimes we fail, sometimes
we try certain things we can.
Oh, let's write a builda service, you know, doing like
fair market access, you know,like or like let's build like,
another exchange, like all theseideas come left and right.
And, long story short,we came up
(18:18):
with a service, solution setwhere our customers like, today.
So which is like we areproviding the most comprehensive
and broadly,you know, utilize infrastructure
for our customers.
And, you.
Know, when you, look at the Web3sort of ecosystem,
as a standalone,what would you say
are the fastestgrowing subsegments within Web3
(18:40):
or just company typesdriving significant adoption on
AWS today? In that category?
I think I really likedwhat you said
about your vantage pointbecause like, it's
very relative,because I work with a certain
set of customers, you know,and these are like in my way,
like, like growing really welland like, like and between
all types of customers,like which ones
(19:00):
I see a lot of activity in,you know, exchange and trading,
you know, and it's,it's very interesting,
like our initial conversation,the move from
traditional finance to, to DeFi,like the this,
this transition is happening,but it is not like
entirely different.
Everything is new.
Everything is different.
(19:22):
It is also following thefootsteps of the proven steps.
Like,I remember one of my customers
said, look,I came with this great idea.
Like where we do the trades,we should do trade compliance,
you know, screening so that likewe don't spend so much time.
I said likeso that is I mean that's
a great idea first of all.
But you didn't invent this,you know.
So I see huge activity and,you know, you can cross-check
(19:44):
what I'm saying with numbersis in, in trading
and this is not like,the jargon in our world is like
the retailpeople are directly interacting
with exchanges.
This is like, this has neverbeen seen for the records
like if you look at the historyof trading, like,
usually it is institutional.
(20:05):
Even if the individualstrade like us,
they are usually tradingthrough brokers.
Some intermediate, companies,and it only happens in certain
hours of the day, you know, incertain places, certain things.
Now fast forward today it is724 retail.
People can literally interactwith largest centralized
(20:26):
and decentralizedexchanges directly.
And also the institutionalplayers are not coming in
because there's a liquidity.
Is there likepeople are interacting.
I'm super excited to and,it may be impacted where I am
because I'm building a lot of,low latency systems, you know,
but it is happening right nowall over the world,
certain parts of the globemore than the others.
(20:48):
And we all know like it'sa little bit eastern way, but,
hopefully in U.S soon and,but the trading is happening.
I think it will fuelother things because once it is,
financially stable world,I expect more, data providers,
I expect more DApps, I expectmore chains doing like,
(21:08):
like day to day, things as well.
And this is separatedfrom what's going on
with the stablecoinsand the tokenization that's also
going where super active.
But I think, in my like, ifyou relook at the growth,
charts, I think I really likethe trading part.
(21:29):
And then, you know,there's this constant debate
in sort of the what'scalled the institutional realm
and their interest in Web3,between different types
of implementations.
Whether that is, you know,blockchain networks or private,
you know,permissioned implementations.
I'm curioushow you would frame that,
you know, sort of thinkingfrom an institutional
perspective and,you know, maybe pros and cons
potentially just sort ofwhere do you see it
(21:51):
playing out over time?
It's interesting.
Again, if we do a correlationwith the past terminology.
Rememberthe first days of the cloud?
There was a big conversationabout public cloud.
Private cloud.
Or I want private,I want public and fast forward.
Like it's so like my my first,you know, interaction with the
the Web3 world or public chains,private chains, enterprise
(22:13):
party, you know, retail likelike all those conversations.
But I think the story willis going at the same direction.
And all my activities, even if,like, institutional,
like large scale enterprise,is currently warming up.
That's the best way to put it.
With the with the vertical,it is all, happening
(22:36):
over, public chains.
There is definitely activityin the private chains,
if you ask me.
I'm in my personal view,like 90, 95,
maybe even more of the activityI'm seeing is in the, in the,
in the publicside of the business.
They are still institutional.
They are still large players,traditional, financial players.
(22:58):
But the transactions are takingplace on the public chains.
Got it.
Now, obviously,like our space, Web3
is really defined by community.
And so, beyond sort of tech,can you describe a little bit
about whatthings AWS does to support
Web3 companies from a businessand an ecosystem perspective
(23:19):
and helping them, you know,sort of collaborate and grow
within, within their own bounds?
It's a good question becauseI meet with a lot of customers,
like companies like that'susually the first question,
how can you help?
Like, are you here toto position on AWS services?
Like, do you want to show mewhat kind of products you have?
You know, that's usually a goodstart of the conversation.
(23:41):
But our answeris very different.
Like we are a part of your team,we are here to grow
your business because it'sa very natural business.
If we are partners in this one,you know, then the growth story,
we we grow together.
It's not like we just AWSgrows more, you grow less.
Technically, you should begrowing way faster than AWS.
So we are actually growingvery healthy.
(24:02):
You know, where you health away.
So in orderto accomplish the growth,
you know we work togetherand it's not just us providing
infrastructure servicesto make you more cost efficient.
And reliable or secure.
You know, it is also aligningyour business goals
and realize them together.
And this involveswith a lot of programs.
(24:23):
We have different programswith different, for companies
at different stagesof their journey.
I mean the startups.
So with the startups,if the moment they get into
the startup ecosystem as a partof an accelerator incubator
or one of the investor,I call to founded or funded,
you know, we directly startinvesting in the startups
(24:47):
and everybody is familiarwith the credits.
We give them, likeliterally tangible, support.
But it is just, like the visiblepart of the iceberg.
We support them with every,every definitive,
like way of like,like giving them
technical resources,giving them blueprints and,
(25:08):
doing joint GTM activitiesand creating opportunities
for themto reach their main audience
using our connectivityor like, like our resources.
We are literallyhaving a full motion
of growing our startupscustomers.
And, this is just oneside of the business.
We also follow the regularAmazon flywheel.
(25:28):
Like,you know, we have a marketplace
or we have a partnershipprogram.
We have the most vibrant partnerprogram, 140,000 partners
in the ecosystem.
It's it's a community itself.
And so,we, we bring in our customers
into this, this ecosystem.
So, so, so, so we grow together,like I heard,
(25:50):
I need to cross check that fact.
Like 40% of the AWSrevenue comes from companies.
We were once a startup, so.
So we are investingin the startups with programs,
with teams, with resourcesat every very different level.
And this is repeating itselfat many different stages
at the enterprise level, fortune500 level, the global level,
(26:10):
we are building dedicated teams,to support our customers.
Very helpful.
And, maybe moving intowhat AWS is doing
and has built in sort of theAI specific, you know, vertical.
You obviously have a very broadand growing, AI portfolio.
Whether that's, you know,SageMaker, bedrock.
(26:32):
Can you walk the audiencethrough how these offerings
align with different types ofcompanies building and AI today?
Yeah,very good question. Because,
we wantto have the most comprehensive,
solutionset where customers can build
and use the technologies.
As I said, theAI technologies have been with
(26:52):
AWS for a very long time.
But if we do a likea like, truly
the main focus is on the AI.
You know, I will startwith that part, you know, so,
we first of all, we providethe most performant and cost
efficient infrastructurefor our customers to build,
you know, to build foundationmodels, deploy them, train them,
(27:16):
and, run their operations.
This is like the compute,the storage, the the well-known
components of AWS is out there,and it's with the, four,
four main layers.
So first of all,we give our customers choice,
you know, a lot of differentchoices.
Nvidia, Nvidiait is AMD, AMD it is
and we have our own processes.
(27:39):
We have our own silicon,we have our own GPUs.
We give very different optionsfor our customers
so that, you know,they can build by their choice,
not because we force themor like they are forced
to do certain decisions.
At thesecond layer, what we do is
as you get into any AI project,the technology is great,
it is creative.
(28:01):
It is like like groundbreaking,but without data
it doesn't work.
So you need to have a very solidplatform to manage your data.
So, what we are doing very hardis that we are also
integrating the data layerwith the AI projects.
(28:22):
And I will explain a bitmore like
like how we differentiatethose, those parts.
But at a very high level,we make sure the data
management, the data lakes,accessing
data storage of the data,processing, labeling,
all those parts are actuallywithin the same ecosystem.
They are all integratedinto your projects.
That's the data part.
And two more pieces actuallycomplement the AWS, logic.
(28:47):
And maybe startshould be everywhere.
But again, and securitybecause security is somehow
when you are super busy,especially with startups
like superexcited about solution like,
oh, let's do itand everybody's on the on board.
But security is hard.
It is.
It requires very specific,very dedicated resources.
So we dedicate, we put a lot ofsecurity resources
(29:09):
and we make sure the securityis managed,
at least with the withthe resource that we provide,
we make it easier.
I'm not saying easy, but easier,better and more secure.
Of course.
And and the last thingthat is the bread and butter of
AWS is scaling.
So how we can scale.
So going back to youroriginal question sorry for
(29:30):
but this is, this is the, the,the managing, logic.
So, what we dois, as I mentioned,
we provide the,the most cost efficient
and performancedriven infrastructure.
Okay.
So this includes the compute.
And we alsoif you want to have a
to have thisinfrastructure managed,
you know I'm buildinga foundational model.
(29:50):
I'm being likelarge inference, farm.
So all those thingswe have a managed product
called SageMaker.
And SageMaker isa family of products actually,
with the SageMaker,we just really is a unified, ID
so, do unified studioactually allows you to access
all these subcomponents,the data layer and the AI layer
at the same time.
(30:13):
And it helps you to,build, train, you know, deploy,
manage, and MLOpsall in under the same umbrella
while managing your data sets.
So that's the SageMakerside of the, solutions.
So this is for companieswho are either directly
accessing,tapping the infrastructure
(30:33):
or building the infrastructureor managing all your, like,
process through a managed layer,which we call SageMaker.
All the notebooks, like all the,all the ops there, they,
they all come with the SageMakerone step further.
If you go, we have the bedrock.
If you are buildingapplications,
(30:53):
you know, you are not buildingfoundational models.
You still do like your own datasets.
You do rac you know, youyou are even like,
like fine tuning certain modelsand you need to access
different models,you know, you need to test them,
you need to optimize them.
And, what we do is,we provide access to multiple
LMS and foundational modelsthrough bedrock, through an API
(31:15):
like that is like, actually,if you build applications once
this is a really great service.
You areyou know, what you want to do.
You want to leveragethe whole technology.
You want to use the technology,you want to build new things
and we have the bedrock layerwhere you have, access
to all this infrastructurewith certain things
(31:37):
I mentioned at the beginningwith the security,
you know, and also,like we have guardrails
so you can control like,like, questions and answers.
And with the agents integration,we will probably touch that
topic later.
And all those, partscome through an API
through bedrock.
So it's easierto build applications,
(31:58):
but it's not there.
We still goin the application layer,
because what if you are notbuilding applications, but
you will still want to leverage,you know, the the technology
we havelocked so many, solutions and
like how your work gets donewith the generative AI.
So, on that layerwe have the cube business.
(32:19):
We you basicallydrop all your corpus,
the data into the solution,and you just start querying
your data sets throughnatural language.
You know, that is doable.
We have two developers.
So you want to build,you turn to your developers
using the assisteddevelopment environment CLI.
Q developer like these are greattools.
(32:40):
It's not limited with thateven from from supply chain to,
call centers.
We built in the AI technologyinto the solution set
so our customerscan use the new technology.
Easily.
So like,like from the user level
to applicationbuilding to the foundation
infrastructure layer,we have three different
sets of products.
(33:01):
You know, I thinkone of the things
that many companiesare going through
the thought process of,and I think you're very well
positioned to maybe opine onwhat are the pros and cons,
or at leasthow to how to think about it.
There are a bunch of differentmethods for building on top
of foundational models, right?
So you have prompt engineering,you have fine tuning of models,
so on and so forth.
How do you think companiesshould be thinking
(33:21):
about the best path to pursue?
As they think about buildingon top of foundational models,
is theresort of a certain framework
you like to go with,or is it much more
sort of company dependent?
One size doesn't fit all.
That's what we learned,you know, and,
it is very dependent onthe resources and the timeline
(33:41):
and the expectations.
This is notjust the, you know, speaking,
you know, and,I've seen our customers
going through this journey,you know, oh, I will increase
my, you know, productivity.
I will increase my insights.
You know, I willI will use generative AI,
but that depends onwhat kind of resources you have,
(34:02):
what kind of timeline you have.
And we learn togetherwith our customers.
You know,usually the path is like
from easier to more difficult,you know, that's that's
one way of I can put things up,like one of them is directly
usingalready built systems, like,
like we have a lot of, you know,solutions, as I mentioned,
like the connect and,you know, the Q business,
(34:23):
like ready to use next day.
You just plug in the chat, the,the the querying.
That all works.
You know, that's one areathat's easiest.
And then you startbringing your own data
sets, the serac, you know,that's usually,
you know, the next step.
You just say, oh, I like it.
But I want to do it securelywith my own data sets, you know,
(34:44):
and suddenlyyou are one step closer
and now you're interactingwith your own data.
In addition to what lumberingzero the foundation model breaks
you, you know, one step down,and you said, okay,
I like this, this lamb, but,you know, it is.
I want to actually finetune this with my own,
you know, theI want the answers this way,
(35:07):
you know, like,I want it differently.
Fine tuning comes next,you know?
So that's, that's requires.
And when you switchfrom actually rag
to fine tuning,I think there's a big step
there, you know,then you really need
to understand the technology waybetter with deeper
and with more,like more technical resources.
Again, with the tool sets,we are there to help,
(35:28):
you know, definitely tryingto make the journey easier
for our customers.
And at the end of the spectrum,we have the builders
of foundational models, whichwe work with them daily as well.
And I think in terms of numbers,I think it's like a funnel.
But, the activity is,almost everywhere,
and it's a learning process.
(35:48):
Some customers just jump inand it's the wrong place,
wrong time.
But it depends on the skill setand the timeline
of our customers.
And and what do you think aboutwhat enterprises are
are doing inAI as it relates to,
the ability to marshal resourcesboth in terms of team size
and data sets.
(36:09):
How is that shifting overthe last 1 to 2 years, meaning
we're seeing massive improvementin sort of the quality of models
and new standardsfor connecting different types
of data sets into models,such as anthropic, MTP.
So do you view it as, you know,these models are getting
so much better that it requiresless people and like sort of a
(36:33):
a smaller critical mass of datato make them useful or is
or is that an assumption?
Just based on, you know,sort of the improvement
of models,you still sort of need
a critical mass of team and datato really get to a place
where value is being created.
Yeah.
In my personal experience,I think, it is going in
two ways.
(36:53):
You are correct,it is getting easier.
The models are gettingmore advanced, you know,
so if we were doingthe same chat bot that we did
like four years ago,it is like literally
light years ahead right now.
You have the chatbot in a few hours,
you know, with your own datayou would like if you use Q
you know, but the expectationsare getting more complex now.
(37:16):
Now our customers are askingfor more now,
like we want to addlike like a very good
question is like, okay,how can I merge the smart
contracts I'm using with myAI agents?
It is a very different questionthen, you know, just,
you know, I want to buildthe chat interface.
(37:38):
You know,I think the time and the effort
is not decreasing,maybe increasing the resources,
but the deliverablesare increasing
in an exponential way.
So there is there is no like,like a linear line of okay,
it's easier.
So I need less resources.
You know, the expectationsare getting higher.
(37:58):
I think the allocationof resources are changing.
I think everybody'sobserving it.
So theallocation of resources to
projects, taggedwith AI are very different.
And this is very normalbecause there is a business
opportunity.
But, my personal opinion is thatthis is not a separate task,
(38:21):
like the security conversationwe had two years ago.
Oh, you should havea security team.
They should do theirsecurity stuff, and we do it.
The business,you know, it is not like that.
So if you are in, like I don't,I don't know,
like in the legal business,if you are practicing law,
you know you areyou have a gene I project
if you are in financial servicesyou have a gene I project.
(38:42):
It is independent of the teamsor like the the vertical.
You're in every business.
This is like integratedin the core of the business.
This is not like a separate teamof AI people.
They are doing great thingsin the lab or in the production.
It has no it is being integratedin the core technology of the
companies and technologiesis becoming a core business.
(39:04):
Like multiple, transitionsare happening at the same time.
Got it.
And you startedto go down this path.
So I think if we canwe should ask this question now,
which is, you know,there's a lot of attention
being paid to agents andautonomous, you know, workflows.
I thinkjust let's start definitionally
because I thinka lot of the audience
may not really know what a truedefinition is of an agent.
(39:26):
So can you give the audiencesort of your definition of like,
what is an agent?
And then in addition to that,maybe some impactful
examples or implementations thatyou've seen of agents today?
Okay.
So yeah, first of all,there is no true definition
because it's just evolvingevery day.
You know, at the end of the day,the agents are small
(39:50):
applicationsdoing tasks for you.
But what definesan agent versus not an agent
is a good conversation.
So the way, I look atthings is like there must be
certain characteristics,you know, in the small
application that we are usingand to be called an agent,
you know,and I think these are like
getting real settled,settling down. Okay.
(40:14):
And, the first one is definitelyneeds to be autonomous.
You know,it needs to be operating
without human supervision.
Okay.
And the second thing,and there's also like
a longer conversational agent,like I was agents because
the agents let's focus on thetopic is actually goal oriented.
(40:34):
They have a task like,you know, to X do y.
So they are goal oriented.
So let's continue.
Autonomous goal oriented.
And the third one isthey're interactive.
This is very interesting.
It is not like an API call.
Go and do this call.
No, it's not likethe agents are interactive.
(40:57):
They get the response basedon the response they answer.
So they are autonomous,goal oriented, interactive. And
the fourth one is they areactually
using their environment.
I read in some place is calledit's a prescriptive.
So they get the weather datafrom left,
they read a file from herethey go to a database
(41:17):
and query a database.
You know, they are perceivingtheir environment,
you know, and they are gettingthe data themselves.
And and theAI part is the last part,
which is they are adaptive,they are learning
and they are changing,you know, so fast forward
whateverI told you, like autonomous,
goal oriented, interactive,perceptive and adaptive.
(41:37):
You know, these are the partsof the small application
which is doing things for me.
So, in order to be called agent,the components are, you know,
first of all, it collects data.
You know, it's on its own.
Okay. It does reasoning,you know, this is key parts
like reasoning.
(41:58):
Not too many people get like,I have a task.
I been runningthis task for years,
but my tasks were doingwhat I told them.
You know, the agents, onthe other hand, are reasoning.
I got this answer.
This is what I'm doing.
And, and also not just reasoninglike my day quickly I'm using
they are taking actions.
Oh I see, and you're using allthese files in the wrong folder.
(42:20):
Can I put them into the rightfolder? Like literally.
And if I say yes, that's veryby way risky move on my laptop.
My, my Shukla is telling thisthings.
You know by the way we have theCLI tells you a lot of things.
It's a very exciting times.
But I'm also very super scared.
So like I say,I will move all your files.
I will create youa GitHub project.
I will create a readme file.
(42:41):
Are you okay with this?
It's like okay, I'm ready,but it's a very different times.
For the things again,like the perception,
the reasoning, the actions,and the last two parts
for the components,they remember things.
It has a memory, you know, andthey are adaptive and learning.
So these are the componentsof the agents.
(43:03):
Those are the small programsthat we are using.
And there is not a single daywhere we do not come up with
new ideas for the last two,three months lately.
So got it.
We'll talk a little bit aboutthe intersection of Web3 and AI.
You know, we're seeing major,major breakthroughs,
you know, globally on unsortedAI specifically
and sort of continuedexperimentation in Web3.
(43:26):
How do you see the AI revolutioninfluencing
or accelerating progressin the Web3 space?
Well, first of all,let's remove the obvious parts
where very quickly,you know, the first one
is the development,which is queue
and sometimes gets very complex.
You know,the development will be easier.
Many people skip this part.
You know,if you have never build with AI,
(43:50):
you know, you don't know the thethe impacts of like the speed
and the the options that theAI assisted code development.
And now I think it was a blockerfor the growth of the ecosystem.
What we are working inis like finding the
the skills and developersand the things that is changing.
(44:11):
Okay, that is very obviousand it's forcing everywhere
to go,but also very specific to Web3.
We are seeing this,speeding up in the development.
Okay.
The projects are comingfaster online
and with more features,more things.
Like, that is very exciting.
Second thing, the most obvioususe case is decentralize
(44:31):
autonomous agents. You know,and there are frameworks
that we can talk about.
There are protocols like,like like like, Ocean protocol,
like the singularity, you know,like there are so many things
the this is very normal.
I have an agent,and this agent is autonomous.
It's self-directed.
It is doing things for me. Okay.
(44:53):
And especiallyin the trading space.
This is this was super exciting.
You know, we have seen plentyof interesting projects there.
And that was very obvious. Okay.
But if you ask me, I mean, I cancount so many different projects
and things, but let me explainwhat's happening in the space.
Okay.
In the spaceI work with the startups.
(45:13):
These are like like we call themengaged, like like startups
really investing in this.
The first step I seeis that, look, all these labs,
when they were trained,you know, they were in trained
with the chain data.
You know, for some reason,I don't know if this is public
data technically. You know,I don't know why.
I don'tknow the reasoning behind it.
(45:34):
So, so suddenly all our peerscame up with the idea.
Look,you could do a lot of things
with the exceptionof the chain data.
So and then, you know,you know, the data providers
and everybody likeand suddenly everybody is asking
the same question, look,we need to fix this first.
We need to make sure thisnatural language interaction
and like, like,like getting all this big
(45:56):
weave of the whole data setsshould include the chain data.
So we are seeing a lotof solutions coming up online,
as like a broker advisorto investor and vice.
So likelike so many different MPs.
So like the, the,the projects that are leveraging
AI technology to work with the,the chain data.
Okay,that's one side of the projects,
(46:18):
you know,and that's super active. Super.
I'm super excited.
I think that will changehow people interact
with the chain,you know, and query the chain.
So that's one end of the things.
Personallythere is a big like I mean,
you can call me the geekor like I'm on the technical.
And so what is making yousuper excited is like
(46:39):
I work with the fintechauto financial service.
And usuallywhen we talk about Gen
they say, look, that is likeif you ask the same question
ten times, ten timesyou get a different answer.
You know, this is actuallyfinancial services.
Not a very exciting thing.
You know, it'sactually very scary.
And the non-deterministic futureof the Gen I, you know, it
didn't work out well in the corebusiness of the fintechs,
(47:01):
you know, but fast forward,what happens if you attach
a smart contractto an agent, okay.
And suddenlythe world is changing, you know,
like now we are talkingabout traceability, like
like absolute transactionson a public ledger.
You know, we are seeingvery different use cases.
(47:22):
So, I think we will seemore and more,
applications on this,a good use cases like,
like we saw with the agents.
Let's continue with the agents.
You know, like, for example,my agent is asking for weather
data, you know,to the other agent, you know, no
humans involved or autonomous,goal oriented.
(47:43):
I want to know the weather,you know?
And the thing is, like,first of all, ownership of the
the confidential dataand the transfer of the data,
that is, that can be done.
And we all knowpeople working on this word
like the, the,the transfer of the data
ownership can be donevery easily with the, you know,
with the Web3 technologiesand one step forward,
(48:07):
I think this is mypersonal expectation and it's,
I think it's happening isnobody's talking about
the financial value exchangewhen all this AI world is
developing, growing every day.
And and I think we will see moreand more projects where I see
I ask for the weather andmy agent says I can spend like
(48:30):
$0.50 today, you know, and I'mwilling to spend 0.1 cents
on the weather dataand like to zero to that.
The Coinbase camewith like like it's an API base.
I think it will soon be likeall on the chain transactions.
That's my personally superexcited, which is oh, okay.
You want to learn the weather?
Just you have to first pay me,you know, and my agent, bound
(48:53):
by the contract says, okay,I have a dollar,
I have $0.50 today.
I could give you a 0.1 cents.
Here is a transaction.
Instant transactions.
You know, almost ish.
We know how fast the chains getand suddenly the transaction
happens.
Money is exchange,and the remote agent tells
my agent it's sunny, you know,and that is the future.
(49:17):
And we see a lot of projects,like there are like, avatars
and like, tokenization and,you know, like,
there are so many different,exciting things.
I think we are, you know,I can speak for the hours,
but we will seemore. Definitely.
We will see more and moresolutions. Great.
Well, you know,we ask all of our guests,
for their high stakes hot take,which, you know, is
(49:38):
what's your boldest predictionfor blockchain, Web3 or AI?
For the rest of this year, 2025?
I mean, prediction or wish it'svery close.
You know, I work with the U.Sbased accounts, so,
I want to see more players in usand, hopefully,
it will be a better environmentfor, for the customers.
(49:59):
And we will seemore, more activity, in us.
So awesome.
Well, with that,I will park it there.
Thank you for joining us.
You know, again, you know Icon.
Thank you so much.
It take forever me absolutely.
Keep exploringand keep innovating.
Learn more at Validation CloudIO and follow us on X
(50:21):
and LinkedIn to stay connectedas we continue to shape Web3
for institutions.
This is high stakeswithout the Walker signing on.
Popular Podcasts
Spooky Podcasts from iHeartRadio
Whether you’re a scaredy-cat or a brave bat, this collection of episodes from iHeartPodcasts will put you in the Halloween spirit. Binge stories, frights, and more that may keep you up at night!
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.