Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Shelby
Recorded: July 23, 2025 Duration: 1:03:39Space Recording
Short Summary
Aptos unveils Shelby, a groundbreaking decentralized storage solution, aiming to transform data management in Web3. With strategic partnerships and a focus on creator monetization, Shelby is set to revolutionize content streaming and data accessibility, promising sub-second load times and competitive pricing.
Full Transcription
Thank you. Thank you. Thank you. No music this time baba
oh man i i think greg is probably more excited than i am you'd have seen me on my seat just
jumping i'm super excited about this space you have no freaking idea um gmgm everyone good to
see everybody's super punchline on this one and i
think we're just getting started with what the numbers of today will be um you know everybody's
excited about the alpha what if i told you guys that the host of this space me and yeah we are
probably more excited that about the alphas than you guys um but yeah we'd like to keep it as something where
we get feedback from listeners as well so please feel free to drop comments in the comment section
and we'll be going through that from time to time probably at a later stage of the space
and um just answering some of those questions i'm super excited about today i mean you can probably
tell by my voice um shout out to everybody that put this together uh i'll tell you guys are super
super awesome and um shout out to greg and praniv that took out their time to come and tell us and
drop all the alphas i mean we're happy to receive the alphas. Let's not forget who is dropping them. But yeah, if you are familiar with the Spooks space,
I think I see a lot of familiar faces on Spice today.
But yeah, if you're familiar with the Spooks space,
there's something we do where we play Pump It Up by Danzel.
It's now the official Spooks theme song.
Yo, man, I see Shelby dropping smileys. But yeah, it's now the official Spooks theme song Yo man, I see she'll be Dropping smileys
It's now the official Spooks theme song
And if you are
Not used to the Spook spaces
I'll tell you, just enjoy
And it's probably
Rub off on you after today
So yeah, Pump It Up by Danzel
Let's go
Let's go. Let's go. Let's go. You got to pump it up, don't you know, pump it up.
You got to pump it up, don't you know, pump it up.
You got to pump it up, don't you know, pump it up.
You got to pump it up, don't you know, pump it up You've got to pump it up, don't you know, pump it up
You've got to pump it up, don't you know, pump it up
You've got to pump it up, it's not so long ago
That's the sound of the nation
Every Saturday night
On your favorite radio
And the puppy's jumping, yeah Every Saturday night on your favorite radio.
The puppy's jumping, yeah.
And the vibe feels so strong.
Now your hands in the air, lift your head up high.
You know you've got to sing along.
Don't you know, pump it up.
You've got to pump it up. Don't you know, pump it up. You've got to pump it up. to sing along
the sound of the nation Every Saturday night
On your favorite radio
The funk is jumping, yeah
And the vibe feels so strong
Throw your hands in the air Lift your head up high
You know you've got to sing along in the air you
what's the name of the jam? Yeah, I can feel it.
You know you can.
I've got my groove on.
And I'm ready to go.
Check out my ride, girl.
But don't catch my radio.
Don't you know, pump it up.
You've got to pump it up.
Don't you know, pump it up.
You've got to pump it up.
Don't you know, pump it up.
You've got to pump it up.
Don't you know, pump it up.
You've got to pump it up.
Don't you know, pump it up. You've got to pump it up. you oh
You've got to pump it up
Don't you know, pump it up
You've got to pump it up
Oh man, I'm super pumped about today
My goodness
I already said shout out to everybody that put this piece together
But I'm going to say it again
Because you guys are dope
But we are used to some faces we are seeing today on the Spookspace
Greg is no stranger to the Spookspaces
I remember there was a period where we no stranger to the spookspaces i remember there was a period
where we always had greg on spookspaces i'm looking forward to seeing more of those anyway
but pranev i'll tell you this the first time i'm probably even speaking to you like on the
afterspace at all um before we even talk about spookspace so i'm really excited about your intro
so i'm passing the mic to Pranav.
I hope I'm pronouncing that correctly.
And then let's start from there.
Yeah, GM, GM. I'm super pumped.
Man, your anthem has got me.
I was sitting down and I'm standing up.
I'm ready to go. This is awesome.
So yeah, quick introduction.
I've been listening to a few of your spaces,
but I'm Pranav. I've been with
Aptos since pretty much the beginning,
and I lead some of the infra teams at Aptos.
But before Aptos, I was at Facebook slash meta,
working on the data infrastructure where I
worked very closely with Avery and
some other team members who are now at Aptos.
And we had built the compute engines
for the entire Facebook infrastructure.
And the number that I like to share with everyone
is that by the time we were done or by the time we left,
the infrastructure was processing 1.5 exabyte of data
every day.
So yeah, there's some interesting scale.
And we'd love to now replicate some of the scale with Shelby.
Yo, talk about amazing numbers.
So yeah, really, really cool.
We would like to be close and personal with a lot of our guests.
Sorry, Greg, I'm not going to do this for you because we know you're ready.
But yeah, Prana, can you tell us a bit about your journey before? Let's talk before you started
working with the team. I know you've been with the team right from the beginning,
but before the team, what was it like for you getting into, I would say, the tech space now?
you're getting into, I would say, the tech space now.
Yeah, yeah, I think that is a, look, it's been a long journey.
Initially, I started my journey with statistical,
building like high-performance statistical model computing.
What that means is before ML became a thing,
statistical modeling and scientific computing was very big. And we're building kind of this infrastructure that
would predict risk management or that would predict risk scores for natural catastrophes
and their ability to impact all the real estate all over the world.
So, for example, like we know there is huge hurricane risk in Florida and earthquake risk
in California and tsunami risk in Japan.
But how do you model that?
So that was terabytes and terabytes of data crunching and predicting and coming up with
probability curves.
That was super interesting.
I did that for first few years of my career.
Then I dabbled into NLP in healthcare,
built NLP engine to read the patient charts
and automatically kind of automate
a bunch of healthcare workflows.
Then I joined Meta, I talked about that.
Then I joined Aptos,
and it was a very interesting journey.
Like year one, we are all looking to head down, deliver mainnet.
I picked up data and analytics.
So if anybody was here during our early incentivized testnet phase,
you know, they have interacted with me a lot.
I was the guy who was counting their kind of their performance ranking and score
and have interacted with a lot of the early community.
Then I've been working closely with the blockchain and research teams on building the blockchain.
Talk about it's the ground running. Bro, you started your career at a really,
I wouldn't call that introductory level. You started your career at a really high level and
I think that's awesome to just see the kind
of minds that we have working together at aptos labs and now talk about you um really contributing
to shelby i think it's super exciting and all the minds included talk about jump as well i think it's
a super team if we can call it, that you guys put together for that.
But let's start from the top about Shelby.
I mean, there are so many, so many storages currently available, decentralized, non-decentralized as well.
Some really cutting-edge technology that we've seen in recent times, because I think over time we've seen technology become better and better especially in the storage space what would you say was the the what would i call it now was the need for shelby what was the problem statement
why was shelby necessary oh yeah absolutely um and i'm happy to take it. But Greg, why don't you take a shot at this?
Oh, man, put me on the spot. Yeah. Yeah. And definitely underplayed, you know, Pranav's introduction to AI and ML in the future. So he is before all that. But yeah, so you want to know why, why is there need for this space and what there's a gap.
So today, if you look at decentralized storage systems, they generally are pretty slow.
And that causes a lot of issues where you might wait a very long time for something to show up, or you might just put something around and not necessarily have the incentivization for someone to provide it as a good way for someone to read it.
If you've ever tried to use some of the existing ones out there, like IPFS, Filecoin, Arweave,
they all have different trade-offs and different paths that basically provide you what is slow storage that you might pay for once or you might have to manage the system for a long period of time.
That actually really doesn't lead to a very good system for building.
to a very good system for building lots of the real-time systems that we have today,
especially for things like streaming or like file hosting and things like that.
Instead, what that mainly is about, like, you know,
maybe if you want to go and store something for a very long time
and keep it for a very long time,
that's more of the purpose that the decentralized storage exists today.
Now, so something like that we probably have seen, especially me,
since I used to work at AWS is hot storage and the ability to view this kind of
information in a faster way is much more important for real time systems,
as well as things as simple as hosting websites.
Because sometimes you want it to be load fast,
or simply where you don't have to wait that long.
So on with it. And you can stop me anytime, Pranav,
if I keep on talking too long,
because I tend to do that, especially on the spook spaces.
Oh, man, no. Well, you have the floor.
You have a guest today. So feel free to yap as much as you want.
I mean, it's Yaptos, he's in onaptos.
But yeah, you know, I find Shelby really, really interesting.
I see Shelby as something really essential
to the Web3 space.
Talk about like all the lapses we currently have. We talk about
decentralization a lot, where a lot of Web3 actually live on centralized platforms, when we
really think about it. And this is majorly due to the last laps we have in storage.
But if this is such an obvious problem, why would you say it has not been solved till today?
Yeah, one simple answer. It's hard.
So, yeah, I'll elaborate.
journey of blockchains and decentralized infrastructure, we spent many years kind of
accepting the fact that decentralized is slow and decentralized means more expensive, right?
And so when you think about, and we are all thinking about this multi-node BFT tolerant
structures where you load data within blockchain and what you can do is like that small data and small compute.
But there is a new store kind of, you know,
all your financial data on blockchain.
To pull this off in a decentralized manner,
you needed a lot of things to come together.
First of all, you need very high bandwidth
to be available in decentralized manner. Second of all, you need to have bandwidth to be available in a decentralized manner.
Second of all, you need to have a lot of infra experience coming together and taking the challenge.
And that is where Jump and kind of Aptos and our background in building high-performance infrastructure,
when we came together, we are kind of ready to pick up this challenge. You need to be extremely performance minded,
extremely kind of infra minded.
You need to think about every byte.
You need to think about your performance in not milliseconds,
but microseconds, right?
And so you need high bandwidth,
you need high performance software,
you need experience to know how your data can be distributed
on all of these decentralized nodes very efficiently.
You need to know how it can be stitched together back
and served out very fast.
And you need to know how your infra will perform under load
and under kind of peak load.
And so when we put it all of that together, now we have, and also you need a very, very
high performance orchestration layer to make this happen.
So Aptos is that orchestration layer.
We have now decentralized bandwidth available.
Like one example is 00 that we are partnered with.
And Aptos and Jump has all the experience that we require to pull this off.
So we put it all together and we were like, well, it's finally about time that we solve this problem for Web3.
Awesome. You know, you started this off by simply saying it's hard.
And to think about a team that says, oh, this is really difficult. Let's do it.
Yeah, absolutely. Absolutely. and to think about it seem that says oh this is really difficult let's do it absolutely absolutely oh man i think that's awesome um i i really like that you guys are
taking on this challenge and it's really exciting i will probably get into um the confidence level
of like taking on this challenge this is something that is uh already like
we've gone through the difficulties we already know exactly how this is going to go i would
really like to know more about that but let's let's take this let's simplify this for an everyday
user because there are so many people that know yes shelby's hyped i mean because everybody's
talking about it so it's definitely hyped But they really do not understand how this impacts them directly.
Very few people understand that there are even different storages.
There's the old storage, and there's the cool storage, and there's a difference between them.
I had to put out an article explaining that.
But yeah, for the everyday user, if you were to describe this to them in a way that impacts them,
how would you describe Shelby for them?
Yeah, for sure. Like every day for everyday user, think about Shelby as the plan that will enable decentralized YouTube, decentralized Netflix.
Netflix, you will be able to see the data, you will be able to follow your creators, you will be able to directly interact and kind of have access to premium content.
You will be able to see the data.
You will be able to follow your creators.
If you are a follower and you have either their tokens, their NFTs, you will be exclusively able to access all of the content.
all of the content and and this is the Shelby enables the creator economy that we have always
talked about and we have been promised but we haven't quite materialized or we haven't been
able to materialize so this is finally the moment where all of this comes together and and for an
everyday user i think that would be the most exciting vision for Shelby that they will be directly
impacted by.
Yeah, and I think that is really awesome.
For the longest time, people have been contributing to so many platforms.
And until recently, you wouldn't even realize how valuable your data was.
I think it took a while for people to understand how much input they were putting to these platforms that they created content on and what it means to them and how they should benefit from it.
Shout out to platforms that really started incentivizing creators.
I think X was one of them.
It took years before people started making any form of return from X.
And everybody just took it as something normal.
And now people are beginning to understand more and more that their data is
valuable and the contents they put out,
they mean something and they should be rewarded for it.
So I think Shelby coming to life is another step in that direction.
And let me add a little bit here,
since you have started touching on value and kind of even extending Greg's initial answer, like why Shelby?
And so Shelby, like we all know, we have all heard this moniker, like data is the digital gold and data is the new oil.
And we have also seen all the big companies have generated so much value from data, like ever since Google kind of built the search engine from there to now.
And the three modes of kind of value creation, like you store the data first,
and then you start serving the data.
So when you serve, you can serve this content streaming,
and you can have website hosting, you have your data APIs.
But when we say Shelby serves,
I think there are two primary targets
for Shareably to Shelby to serve data.
One, serving the data to end users,
and that's where we talked about the content creators
and everything else.
And secondly, serve high throughput data
to data processors, which could be analytics, which could be model training.
And that is where kind of data gets converted into signal, right?
That is where your data gets converted into intelligence.
And the value unlock from data becomes exponential.
And this entire layer of being able to serve high-performance data
for either serving the data to users or serving the data to
data processors that has been completely lacking. So Shelby, at its very core, Shelby is the value
creation engine that we are envisioning. And this is where users like content creators will be able
to get more value from their content. They'll be able to monetize it in many different ways.
There are different models coming up for AI now where your data, your personal data is very, very important.
We have a partner, Flashback Labs.
They want to take and anonymize your data and put it on kind of data marketplaces so that a user can directly monetize
all the data they have in a very anonymized fashion.
And this is where the models become much more smarter
and then you directly get paid
for contributing your data to the AI economy.
Yeah, and talk about like inclusivity.
Now you can really feel like you're
part of those platforms that you contribute to,
and you benefit directly from the platform benefits.
So, Shelby is designed for
read-intensive use cases like video streaming and AI inference.
But earlier on this piece,
we spoke about how they are decentralized storage platforms,
but a lot of them are really slow.
And I would say one of the reasons why they are really slow
is because they are trying to be decentralized.
I think that's where a lot of the problem really lies.
So how we shall be able to achieve this speed
and still be decentralized?
Yeah, yeah, great question.
And I think there are kind of six pillars we have identified in our white paper, and I'll summarize them.
One is that all the nodes, like if you look at current decentralized storage infrastructure, the nodes go over public internet.
Public internet is congested,
they don't have enough bandwidth,
your nodes are distributed all over
the world with different varying bandwidths.
How your data gets distributed across all of these nodes,
and when you want to read the data,
you have to stitch it all back together,
or retrieve it from multiple nodes.
And just going over public Internet slows this down.
So Shelby is going to use dedicated fiber network.
So that is going to be 100 Gbps or more network bandwidth,
completely dedicated for Shelby nodes to talk to each other.
And this network is also going to be decentralized,
by the way, so it's not a centralized network.
It's a decentralized network.
There are lots of fibers that is laid all over the world
and they are dark right now.
They're not being used.
And Shelby aims to tap into that unused
and header-to-unavailable network bandwidth.
And double zero is a partner
that's going to help make that a reality.
So that's part one.
Part two is we have come up with novel auditing protocol.
And we can go into more details if required,
but the challenge with decentralized storage
is that you always have to audit your data
and make sure data is available and it's robust.
So we have come up with a new auditing design
that reduces overhead of auditing
that many systems have to take and kind of
completely almost eliminates it. The third is as I mentioned we need a very
fast coordination layer and we all know how fast is Aptos it's like almost like
now 500 millisecond latency so any kind of coordination required to write new
files becomes very fast with Aptos. Aptos after scale also we are familiar with that
so we can also support very high throughput and very high peaks we have something called data
encoding scheme which is you know a lot of these protocols kind of take multi x times replication
overhead so i'm going to take your data and replicate it like 10 times or five five to ten also kind of take multi X times replication overhead.
So I'm going to take your data and replicate it like 10 times,
or five to 10 times all across the network,
so that even a few nodes go away,
we don't have to worry about losing your data.
And Shelby takes a route of getting encoding scheme,
which is called, the scheme we have chosen is ClayCodes,
which gives us this optimum performance
between kind of storage overhead and bandwidth overhead.
So that is, and then on top of that,
there is going to be very high performance software
that is going to power the entire network.
So if you try to access a file through an API,
it kind of goes through many layers of operating system
and buffers, and that process itself is very inefficient.
And Jump has this internal storage system
that they're built where they bypass bunch of these layers
and directly kind of from disk to serving out the data
that is very minimal overhead.
And we intend to leverage a lot of that already that exists so it's battle
hardened software uh efficient encoding scheme novel data auditing scheme high performance
coordination layer and dedicated fiber like all of this comes together to make shelby this fast
yo greg seems like greg wants to add something but i wanted to impute here that you said
high performance software powers shelby and i'm like oh do you mean aptos because it definitely
helps that shelby's building on aptos for greg go ahead yeah i mean i wanted to just put one more
color on this point and make sure you guys think about it. A lot of these high performance systems that provide for things to be sent to you really
quickly including groups like Facebook and things.
How do you get something that is served to you as quickly as possible can only come from
people who have experience and to do these novel changes and these novel things
to be able to make it work.
There's something specific even at Facebook to load your Facebook page as quickly as
possible from disks.
And that kind of experience that both Pranav has from the infrastructure side as well
as the jump side is the way that we can make it so that you can have it decentralized in a way that is quick to you as a user,
but doesn't necessarily have the same downfalls as most decentralized services.
Yeah, and I think this is where the experience of the team really comes into play.
There's so much from, like, I would say experience, once again again that the team has seen and we are now working together
to merge this to create this amazing product that is really going to just revolution revolutionize
the whole um web3 space but um i had a friend read through the shelby white paper and he said
something that stood out to him that still
doesn't understand how it would work is that Shelby is going to charge for reads.
Why is this even necessary? Let's start from there. Yeah, my favorite quote here is that
show me the incentives and I'll show you the outcomes. And I'll unpack that.
We are incentivizing reads so that every storage node will get paid for serving the data at high performance.
And so if they are not going to serve it, somebody else will serve it and they will get the reward. So first of all, that's the incentives for the node operators and
node storage providers to give the performance to the network.
Second of all, if you pay for the reads,
you start storing data that is worth paying for, right?
So I think this is a very powerful statement. You, we have seen like a lot of storage systems
where there's, hey, just come and store the data.
And there's also like something like store to earn
and some of those schemes.
And then you collect the data that just sits there forever.
Nobody knows what's the value.
And you just have this massive storage overhead
and there is no real value being created.
When you say you have to pay for it,
and let's be very clear,
we are all paying for high value data right now.
Whenever we are streaming anything that is worthwhile,
or whenever we are processing data that is worthwhile,
we pay for it.
And that is how value gets created from data.
And so I think this read incentives are extremely important for Shelby.
It's actually very fundamental to what we are trying to do.
First of all, it incentivizes storage providers to serve data at high performance and
keeps the entire system high performance.
And second of all, we kind of start attracting data that is worth paying for.
And we have this whole cost model that we are putting together where we believe that the high read traffic and with sufficient number of reads, in fact, storing the data and reading the data from Shelby will be actually cheaper than current cloud providers
and current big names in the cloud.
So that is, again, a very exciting kind of economic model
that we are trying to introduce.
Oh, wow.
Okay, definitely if it's cheaper, it takes a bit of the worry away.
But I think about this,
probably it's just me looking at it from a wrong point,
but I think about this as an extra layer
that we can probably do without
and is going to assist speed.
Because I think of it as every extra layer
of getting this data is going to slow down
how fast this data is being, would I say, posted or
saved. Would you say in any way this charging for reads is going to impact how fast this data is
saved? Ah, great question. No, absolutely not. And Shelby's read optimized. And so we have removed
all inefficiencies from our read path.
And so this whole payment mechanism
between kind of the storage providers
and the RPC nodes that actually serve out the data
will be handled through payment channels,
which would be kind of somewhat asynchronous
to the actual serving out the data,
the flow of serving out the data.
The payment channels asynchronously keeps track of all the payments,
and all the reads that have happened,
but the actual throughput of reads will not be impacted at all.
Very good question. No impact on read performance.
Awesome, awesome, awesome. Good to know.
I would really just like to see a lot of this in practical.
I feel like we are so excited
and everybody wants to be a part of this.
I think this is one of the reasons
why I felt it was important we had Greg here
as Greg is now head of DevRel.
You know, I've not been able to tell Greg
congratulations on that role
because I almost feel like it's nothing new.
He's been doing the work way, way, like a long time ago.
Greg has been my go-to person for everything, like, Dev-inclined,
and he has been so awesome at it.
Thank you, Greg, for being such an awesome individual on the Aptos space.
And I wouldn't still say congratulations on the road.
But yeah, there are so many people
looking to build into this.
Remember I said my friend read the white paper.
I was looking at how he could be a part of this.
And that's where he saw the pay for reads.
And so, but I would, from your point of view,
if you were to advise people looking towards building something on Shelby,
how would you introduce them to it?
What should be their entry point?
Oh, Greg, that's for you.
Oh, sorry.
I know.
The muting, I always forget to hit the button.
So from the interim point of being able to develop on Shelby,
so at the moment, there is not a network directly to test against.
However, you can sign up on the Shelby website for information about how to get connected
on the preview that will be coming up soon.
And additionally, like just think about the parts
that you can actually think about
how you're gonna interact with it.
So thinking about how you can,
learning about how to interact with the Aptos DevNet and be able to build things on top of it, there's a Shelby CLI and SDK associated with it.
coin. It's going to be seamless with some bridges that allow you to be able to connect the existing
things that exist today for stored providers and look at it as well. There were some people
actually asked me about, hey, can I make this look like S3 and similarly? So it's going to be
probably similar kind of experience for being able to store kind of information and just you just need to sort of
get started learning about the parts dealing with move contracts the price logic and the parts that
interact between aptos and shelby as a whole oh i think you you okay pranav please go ahead
yeah no just a couple of things on top of what Greg mentioned.
We have an early version of CLI and SDK.
And some of our, you know, the launch partners who the names you have seen on Shelby website,
some of the partners are actually already actively trying it out.
Very, very soon, in a matter of days, you are going to see an update on our Shelby website where we are going to open up a form for the community
to say, hey, I have some ideas.
I would love to either build some utility on top of Shelby
or I have some product ideas that I want to build.
And we are going to open it up to the community
and say, hey, if you have some ideas, list down your idea.
And we are going to give you an early exclusive access
to our SDK and CLI that's hitting a Shelby server
that's hosted privately for now.
And in couple of months,
we are going to have around Aptos Experience timeline,
we are going to have Shelby DevNet,
which will be public facing,
and everybody will be able to start building against it.
So that's the overall kind of approach.
If you are looking to build on Shelby, start thinking about these two upcoming milestones.
Yo, Franav, you just dropped Alpha.
Have you said this anywhere before today?
Not really.
This is the first time I'm saying it saying oh my goodness
okay it definitely pays to be on his book space we somehow fish fish out these alphas from our
guests well yeah franif thank you so much for that i mean so many people anticipating this
and um you just dropped gold now uh and i like the feeling that we don't have to wait so much um soon everybody
will be able to see what this looks like um give reviews as well and yeah just the community can
announce that the community has been involved anyway but the community can get more involved
in this um greg it's a bit of my next question by the way but let me let me tell you something
about greg because you introduced and you said you know it's a head of my next question. By the way, but let me tell you something about Greg, because you introduced and you said, you know, it's head of DevRel, but he has been playing that role.
But, and Greg knows this, when I joined Aptos, and like I had, like it's a, you know, and Elvin is a pretty deep stack, and a lot of things are going on technically.
And I used to tell Greg that he's the guy who knows everything about everything.
Like if I had a question,
and if I cannot get answer anywhere else,
I'll go to Greg and like,
hey Greg, how does this work?
And it doesn't matter which part of the stack it's at.
It could be about anything about key management,
even some cryptography or a blockchain works,
how networking works, how our standards work.
So yeah, I think Greg's been playing that role
even internally for quite some time.
yeah, it's just a matter
of time that we just
there was a formal title with
this role.
This is exciting.
It's very true, very true.
I feel like I've been playing that role
for a long time.
But I did want to shout out that it's nothing without the community of you guys, whether it be the Spooks guys, Baba and Yarn, running these spaces, whether it be Alex and Angel and even some of the more consistent community members who have especially come on the spaces that i've done like i mean i think
irvin down there i see he jumped on one of the spaces once upon a time i remember he was pitching
to me his uh mayhem game way back when like i don't know it's two years ago on the aptos
spaces and then like other people like kiko Chu, TTV, Ivan,
even some newer ones like Handle and people.
And it's nothing without you guys in the community.
I love having you guys.
And it's only if I can help support you guys in the space can we have a community
and build such great products for you guys.
Yo, awesome, awesome. such a legend man um you
know really greg knows a bit about everything talk about yapping there's a yapping leader board and
i think he probably decided he was just going to cool off because at the point greg was dominating
that i was about to enter greg's dms for tips so maybe i'll still do that anyway greg you're going to
give me yapping tips so now i can become a top yapper as well since we are all going to greg for
all our questions but but yeah um earlier you spoke about a bit on migration because it just
makes total sense something as awesome as shelby cannot come into play and i'm going to have my
data still on i wouldn't want to call them redundant storages but still on platforms that
can serve me as well as shelby so is this a priority for shelby to make sure people can
migrate data from other existing storage platforms to Shelby as well.
Yeah, migrating existing storage platforms to Shelby.
So I blanked out, kind of like, you know, Twitter rugged a bit on me.
Yeah, so maybe I can take this.
So look, I think absolutely that is going to be in our path of productionization. That is absolutely going to be a very helpful utility to provide all of our builders. But at the same time, this is also a great opportunity for somebody in the community to put their hand up and say, hey, how about I build out this utility for the community?
and say, hey, how would I build out this utility for the community?
So we are looking forward to builders as well to say,
how about I build a plugin for Shelby with this X storage protocol?
So if you're interested in doing that, reach out to us.
And as I mentioned, in a few days, we are going to have a Shelby website.
We'll have that form.
And please express interest.
And we would love for you to build out more of this
utility for the community awesome man um i'm really looking forward to see how people begin
to migrate i know so many people that are waiting for this so many things have been stored in so
many centralized platforms and i think it will be beneficial that everybody can now have this
or enjoy this decentralized data that still moves really fast. So let's talk a bit about creators.
I know this is the creator season in the whole of Web3. Shout out to Kaito for making that happen.
I think more than ever, creators have been rewarded in these times.
And it's really exciting to see how Shelby is going to play its role in terms of content creation.
I've joked about Shelby being the next Netflix, being the next TikTok.
Like, it just comes on and on and on and on when we think about what Shelby storage can be.
But when we think about it in depth, is Shelby going to be all these things at once?
Or would they need partners that are going to build this infrastructure?
For example, are we going to be watching videos directly off of Shelby or it has to be off of a partner using shelby yeah i again an amazing question great
question so um video serving and content serving um especially viral content uh serving requires
presence as close to your device as possible right so? So we would love to have presence of Shelby content
being served from as close to the users as possible.
Shelby itself will have global footprint.
Shelby itself will have this RPC serving nodes,
and the serving nodes will be also free to cache some data.
And, but some partners are already excited
about being just the serving layer,
not the storage layer.
And so we have you know somebody
likes a pipe network from which
actually has built on Solana,
but they are excited to use Shelby
to serve out some of its content already.
And they have like lot of hyperlocal presence.
We have couple of other partners
who have reached out.
We have some partners who want to
build custom RPC node
to serve out certain type of data
that they are seeing demand for.
So Shelby is kind of going to be this global layer
that makes data available extremely fast.
And from there, if you want to serve out
to like a particular geography or particular content,
there could be an additional layer that serves out the data.
So the answer is it's going to be a combination of both.
We are going to have some custom layers
that are very focused on serving out some custom content,
and there is some data will be directly read from Shelby.
Oh, okay.
I think that is really cool
because I think about it as if you wait for front-end
partners sometimes they won't maybe do things exactly or to the best of how you would like it
being done and now talking about all the amazing things we know can come from shelby shelby
it would possibly be every and anything um throwing this to Pranav and Greg, I would like to know personally to you, what are you looking forward to the most?
What kind of platform are you looking forward to the most being built on Shelby?
Pranav, let's hear you first.
Yeah, I think there are two big kind of use case buckets.
One is going to be in content streaming.
I was looking at some stats and on internet,
currently there is kind of 500 zettabyte of data
is being created per year,
and 50% of that data is video.
And if you look at even video and some sort of,
between video and audio, like look at this,
you know, the spaces, the space that we are on, or look at our overall digital content consumption,
video and audio is dominating the time we spend on devices. And so just being able to serve out content very efficiently being
able to provide more monetization models being able to provide some sort of censorship resistance
being able to free up the content from all of these monopolies that these platforms have created
i think that is going to be very very big so that's one whole kind of bucket and there are
many different ways content can be created served out and monetized and we see one whole kind of bucket and there are many different ways content can be created,
served out and monetized and we see a whole suite of applications being built.
And then that is the second big bucket is going to be, you know, getting converting data into intelligence, right?
So that is the AI use case.
And that's where various, you's where AI data marketplaces,
we have a partner who specializes
into nature photography and nature videography.
And they are being inundated by all of these AI engines
to say, hey, can you give me the specialized data?
Of like one example is like,
can you give me this high resolution shots of
polar bears from many different angles?
And all of that data is extremely valuable and they want to
put it on the marketplace and sell the data for AI training.
Then we have a partner who wants to create model benchmarking and
run all sorts of models with different use cases.
For example, which of this existing LLM model
is most efficient in working with spreadsheets?
And they want to create benchmarks,
all sorts of benchmarks.
And so they want to use Shelby for hosting data
and hosting all the outputs from the models.
So that's their use case.
I talked about the partner that wants to
have user data being sold on marketplaces.
And the last, but not the least,
is the whole augmented generation category of AI.
So streaming and AI are going to be
this two huge buckets of use case,
two huge categories of use case, and Shelby's going to be this two huge buckets of use case two huge categories of use case and Shelby is going
to be very focused on making both of them successful yeah let's go um Pranav is definitely
excited about building Shelby we asked for one use case and he gave us multiple oh man that's
super exciting like we every single day we are getting pinged
by somebody with another exciting idea so so we are we are stoked you just wait till devnets bro
i i can imagine some submissions that are going to be there like people joke around the silliest
stuff but i'm sure people are going to put this forward as well. Okay, I'm not going to say those things,
but I think if you think about it a bit,
you would know we are going to.
Well, Greg, I would like to see your most exciting use case for now.
Man, he covered so many things.
I'm trying to think.
You know, I feel like every time I talk about use cases that I always want to see, I always talk about gaming.
So I'm not going to talk about that.
So I'll say, like, I think for me, it would be super exciting.
I think also in the photography use case, as an amateur photographer, I have always run into all my limits for as cheap as I'm trying to pay for my Google Drive.
for as cheap as I'm trying to pay for my Google Drive.
And the ability to be able to train things,
to be able to say like,
hey, I want to be able to index and search this information
so I can see, you know,
the most common use case is to show pictures of my dog.
Everyone always wants to see my dog.
So I have to go and find in my Google Photos
all the cute pictures of my dog and put it on.
And even being able to do that or train models to make even cuter AI-generated pictures of dogs and based on that.
And I can sell the IP of the fact that these are pictures of my dog, which would be super cool in both AI generating
something that is the future, being able to create, you know, cute little characters,
and then also, you know, show off my dog.
Yo, I think that is really sweet.
And you just made me remember when like photography used to be a thing on OpenSea.
I still think it is because those communities were formed
and then they were like these photographers
growing right out of these markets
that were coming for photography in the decentralized space.
I think we're most likely going to see something similar to that
when Shelby goes live as well.
We are creators can form communities around Shelby
and get paid for the work they put out
i mean just think about like a photographic community being able to monetize their photos
on shelby and wow there's so many use cases
i think creators have always been underpaid for the work that they do. I think one of the things that I used to always hear when I was at Facebook was that Instagram is, I think, the worst platform for monetization.
Even though it's definitely a place where you can post your pictures and you can get a lot of hype and excitement,
if you look at a lot of those platforms, what they'll do is they'll actually refer you to other platforms so they can monetize off of it. And in fact, being decentralized, you can own it directly,
you'll be able to probably be able to have a platform that lets you ensure you own the
monetization that's associated with the content that you create, rather than having to work across
different platforms just because, oh, this is a better place to get people
and this is a better place to make money off of it.
And keeping it all together in one place
is definitely the way to go.
Yo, definitely, definitely.
So I also used to be a photographer
and I remember hoping through different platforms
because everyone offered like a different aspect of benefits to you.
And you kind of had to fine tune your stuff for this place and then realize, OK, on this place, people wanted something else.
Well, man, think about putting everything together and just making those photographers' lives easier.
Let's talk a bit about real world use cases one of the most exciting things again for me
is that um shelby has been compared to aws oh and i think greg is such an amazing person to
have here when we talk about this okay so shelby has been compared to aws and i think that is
really really um would i, such a high target standard
that we already have in our minds for what Shelby can be.
And are we really looking at Shelby being able to offer
web two-level streaming, like real 4K videos
from a decentralized backend?
Yeah. I mean, like the whole point of this is to be able to provide that.
So the benchmarks that we're looking at for Shelby is looking at sub second start times for a stream. So being able to do the full like, you know, you go to Netflix or whatever, and you try to watch your stream, sub second to load. And then also, the costs comparable to Amazon S3 on storage.
So it should be roughly the same or in some cases cheaper, as I think Pranav mentioned earlier, and the ability to have that consistent high throughput performance.
The funny thing is, I think it actually may be much faster than sometimes even like S3 in the distribution, depending on how people are using the amazon cloud services because
a lot of times and i remember from working there you have to set it up so you use cdns
for the different regions that you're in setting up that you make sure that the data is hosted
wherever you were like companies out of brazil would actually host it out of north virginia
but the thing is the time that it takes for the content to be hosted from North Virginia to
Brazil is a lot longer than, say, if Shelby had a node in Brazil and it has a direct fiber
there, it will be much faster to be hosted there.
And you won't have to necessarily manage that confusing part where you actually have to pay for infrastructure in each different
like region and then all the extra optimizations that you have to do there oh my greg just said
shelby is going to be the fastest storage platform in the world that's right. Oh my. Yo, let's go.
Okay.
Definitely alphas have been served.
And everybody suffices.
Here's one more alpha.
So we know that Shelby,
kind of the entire coordination layer is Aptos,
but it is cross-chain compatible.
And when we announced,
we announced that Story Protocol will be using Shelby,
Metaplex will be using Shelby,
and Pipe Network will be using Shelby.
And I will not share the name,
but one another top tier L1 has reached out and they say,
hey, there is a lot of data that we see
being requested and stored,
and we'd love to use Shelby as a hot storage layer for all the data that data needs that we see being requested and stored and we'd love to use shelby as a host storage layer
for all the data that that data needs that we are seeing so that's another big partner coming
pretty soon and so yeah we are super stoked super excited oh my okay i think the rush is just getting
started and you probably even get a lot more shout out to the b BD team of Shelby. I needed to say that.
I was going to say that earlier on this piece.
But shout out to the BD team.
You guys have been hard at work.
I love the fact that when Shelby was announced,
it was announced alongside several partners.
I think that just shows how much you guys have been able
to make people believe in this.
And more and more people are believing more
in what Shelby is going
to achieve and I think I personally just love the feeling that we don't we don't seem too far away
but to now talk timelines I know a lot of people would want to hear this you've spoken a bit about
DevNet but what is it like to mainnet?
I think, why don't we save it for our next basis?
Let's drop alpha one at a time.
But I'll tell you this,
we can't wait to put Shelby in the hands of developers.
We are super excited with all of the response we have got,
and we are looking at trying to optimize all the timelines possible.
So let's wait for the next phase.
And then that's where we can share more timelines after DevNet.
Yeah, let's go.
You guys have been hard at work.
I think that is really awesome.
You know, one use case I'm really excited about is
as much as Web3 individuals complain that they spend all their life
on X and on DeFi protocols,
maybe soon you guys will be able to get girlfriends as well off of Shelby.
Put those gracky girlfriends on Shelby.
You're bullish on Shelby. you just got the new girlfriend
okay that's going to be awesome and i think you guys have been super exciting on this space
learned a lot about shelby personally can't wait to see more from it but going into the future if
you were to i mean people are already excited and bullish about shelby but if you were to, I mean, people are already excited and bullish about Shelby.
But if you were to add a bit of bullishness to that that already exists, what would you say people should look forward to the most?
Oh, man, there are so many things.
Number one, one of the biggest Hollywood studios
is building a streaming platform with Shelby.
Number two, we are going to have a complete creator economy,
creator tokens, creator incentives
that entire platform is coming to Shelby.
And we should do a separate space.
And Andrew from Aptos is digging deep here
and so we should bring him on
and we should do a completely different space
and just deep dive into this new way
of consuming content through top tier creators.
And we are expecting some of the big names to join Shelby
and put their content there.
And the third, I think we'll save it for later,
because again, that requires a lot of time and space,
to explain is all the AI
use cases so yeah I think there is just two or three big massive things we are going to announce
at Aptos experience so again keep stay tuned for more alpha that's coming and I think that's what I'm going to stop sharing now.
I might end up sharing too much.
Yo, I love it.
So earlier on in the space, I said people should drop their comments in the comment section of the space.
I don't see a lot of questions, but I see so many praises for Shelby and what is going on.
A lot is going on under the hood.
For Shelby, I mean um pump it up okay that's
the spooks theme song but yeah just so many positive comments shout out to everybody that
came on for this one I saw a really cool one from Mr. Zarok as well he said somebody should build
blockchain analytics platform Mr. Zarok puts a lot of analytics about Aptos and Aptos DeFi
platforms as well. I think that has been really useful for me myself. I think it would be cool
if we see that on Shelby. I mean, there's so many use cases. I can't begin to imagine what we are
going to see come from Shelby. And I really just think about how this is going to be organized but it's something i would
like to see practically than being told about but yeah thank you everybody for coming to this one
it was alpha packed glad to force out alphas from greg and franev uh to our guests as well that came
to listening you guys were awesome and feel free to put out your feedback
or questions i'm sure we have a team that is ready to answer any questions as well if you are looking
to get involved with shelby trust me greg is going to answer all your questions he already did that
for years now but yeah um thank you guys once again for coming next week we're going to have another spook space but yes
that's going to be on tuesday now this is a special space next week's space is going to be
back to tuesday 4 p.m utc and i can't wait to see you guys on that one as well shout out to every
angel uh brent the man and all the amazing people that came into listening as well.
You guys are super duper awesome.
And for now, it's bye from me.
See you guys next week.
Love everybody.
