Building the Shared Computer on Solana with Bless Network

Transcript

we call it the embedded node what it means is for the future applications to be user powerered you're just users of the application and you're automatically powering and hosting the next user that uh comes online to use the application and I think this is really the ultimate the end game right where we envision the new internet uh should be you know especially as AI starts to move off of gpus uh and back into CPUs um we don't want to really restrict ourselves to the actual underlying technology uh we prefer that where the most Supply is we can bring the demand and reduce the [Music]

cost I'm here with bushan Lee and drick Anderson their co-founders at bless uh which recently rebranded from blockless you may remember botown from a previous episode about a year and a half ago uh when she came on the podcast to talk about uh the blocklist vision and a lot has happened since then the protocol has rebranded the bless uh really narrowed and focused their product vision and is now live and there's lots of exciting stuff uh coming to bless so I wanted to get them back on to talk about what's new in the BL uh product ecosystem and exciting things ahead so thanks guys for coming back on awesome glad to be here thanks for having us thank you y yep so

I want to talk about this Rebrand because I think it's like really one of the most exciting things that's happened in the last little while uh also the Chrome extension uh is really cool I've been using it I'm running here on my Mac that might be why my computer is uh is uh you know the fans are going off as as I'm also doing podcasting but uh yeah talk a little bit about what the vision is here like and what was the motivation behind this Rebrand and this um this Vision change this focusing of the vision on this idea of a decentralized shared computer bless is um the Rebrand of blockless and we're all about making it accessible um for everyday users to get

board on this new economy and this new world that we previously call it web 3 but these days I believe this is just the open internet a new internet that um we are building that include um everyday user as both on the consumer end and also on the supply end of the compute resources so uh bless is um just a short version It's simpler it's more accessible for uh the people that are in the ecosystem and um se you mentioned that we're building this world's first shared computer right so this is a grand vision of how we're leveraging those consumer

devices like laptops mobile phones um tablets um that uh don't uh they don't just access the internet but they actively and collectively power the internet and with this we unlock this full decentralized um faster and more efficient um big scale of the U compute infrastructure the ash network uh and it dynamically scales uh with each application's own user base and in real time and it's all done automatically behind the scene yeah I'll just throw in you know I think when we started the extension

Journey we originally had envisioned probably a product that was not that um we had started ideating with a way to monetize and reward operators uh in a more streamlined fashion by actually powering portions of the community so you know today in in traditional web 3 or I should say blockchain your operators are really just validators and they're validating transactions and then kind of in web3 infrastructure they may be supporting some kind of dap um but but really little are they actually interacting with the community so we started with what would happen if the website or the web page itself was a node and that node then connected to the rest of the network and it really just

organically spun out of you know maybe we could reduce all of operator onboarding um you know at least residential and real um excuse me and uh in general purpose uh you know uh retail there we go that's the word I'm looking for the retail Community um we could onboard them really fast almost instantly by providing a zero friction and really then it just kind of you know why not just making an extension that you can install into one of these compatible browsers and then now they're instantly running a node uh within the network uh especially since we use WM so they can participate in all general purpose uh compute on the network itself yeah I'm very pleased to be also

participating in this uh compute Network I've got the extension running here on my uh on my on my laptop for the last couple of weeks and uh seeing that I'm contributing to to the Blessed Network it's really exciting it feels you know it feels a lot like running a Bitcoin node in the early days you know it feels like you're contributing something to like this kind of Greater resource um are there any plans for for the bless extension or I don't know if it's already the case like for people to be able to run this on their mobile phones or or even like on server infrastructure and I guess you know uh another question that kind of ties into that is do you expect bless to remain this sort of computer consumer or retail focused sort

of product on the on the uh on the supply side or do you think that at some point bless will also professionalize and attract professional node operators professional infrastructure providers that may have just excess Supply and want to contribute that to the network yeah so we already have uh app running on some people's mobile phones it's just because those uh devices their battery constrained so um we um the system is mindful of those constraints and currently now um providing the um those those phones the um capability of connecting to the

internet over a course of certain time uh for the next couple version of the release we'll have the client appli for say desktop version um and also mobile apps um to make it optimized for each of those local devices um so um short answer is absolutely um it you can U imagine it running on any kind of consumer uh devices as long as you can uh open a web browser on um on those devices a no to the institutional participations uh yes we actually have um over a 100 uh Institution professional grade um Hardwares servers um participating in the network and

that's kind of the beauty of the tax stack uh we have this automatic orchestration that does the matchmaking between the um workload that comes into the network and the device uh that are running those workloads based on the device capabilities availabilities and all that and I'm sure Derek can explain more on the technical uh details how it's made possible but that's really our uh bread and butter for even being able to kind of run this Uber like business for uh the compu resource uh matchmaking yeah I'll just throw in uh you know we designed the base infrastructure and the

protocol to be readily available for all low powerered uh devices even those that I would not claim as consumer grade but may be industrial but could be running on the edge uh as an example um you know arm 11 devices 256 makes of ram um we do Envision that that these nodes could go anywhere uh work needs to be done um and and we do Envision that that we'll have this nice niche between you know uh retail being in the middle and then high-end institutional uh on one side doing maybe some kind of crazy AI inference or ml or some some big

audacious thing but at the same time there's a plethora low power devices that are doing things like face detection object detection you know sitting trying to do as much serverless as possible so that they can do some kind of detection before it actually needs to be brought into the middle uh for further analysis so we also Envision being able to run on those devices too okay so basically the if I understand correctly the type of workload is getting deployed or um yeah the the compute resources are matched with the with the right type of workload dependent on the weight and uh and demand for that workload so if I'm just running some web page that might be

distributed across some nodes that are running the browser extension where you don't need a whole lot of Computer Resources to do that but if you want to do some AI inference that's going to be deployed to some much more powerful set of nodes that can manage that that that workload yeah that's absolutely correct we enumerate capabilities of the node and then those capabilities are advertised and then they're rated and ranked based on their ability to perform their adverti capabilities right and so how do you then um like how does blocklist know when I'm deploying some payload as a developer I think we'll get into this also like you was a developer experience but you know as a developer when I'm de

building some payload or sending it to to blocklist how does blocklist know what is the kind of like demand a compute demand that this program will um have on the network and and what type of node it will need to be deployed to so uh you know the developer develops uh their assembly and their package and that gets uh uploaded to something like ipfs along with that deployment to a uh storage chain so that we have you know that automatic fail over and redundancy and all that goodness that we get from distribution um they send a manifest that dictates what the package is looking for and it's you know it

typically at the benefit of the developer to make sure that their manifest uh is appropriate for the app because they'll be charged based on that right yep and then we do um before we actually do an execution it if it feels like it's a really long process to talk about but it literally happens in in milliseconds we do a simulated analing uh across the nodes that would be coming in and we digest which ones are more likely to return the proper results not that we just favor those because we feel like part of what's missing in web 3 is this automatic distribution and actually doing operator filling so you know if you're sitting there running the node on your browser and you never get any jobs like there's no incentive for you to run

the node so we want to make sure that we're not just always waiting towards you know the fastest most performant we really want all honest nodes to be able to participate okay so this manifest I it's you I've deployed some applications like AWS uh using the command line tool and there's some Json file where I'm specifying to AWS I want you know a four core and this much RAM and this much kind of hard drive space and then that you know my application will get deployed to that type of server that's similar to the Manifest here where the application developer would specify what he thinks the application will need uh otherwise the application

you know won't run or won't run performant that is that kind of the way it works yeah that's that's exactly it uh we also do have uh you know a visual online uh kind of portal that somebody can interact with and change you know what those settings would be as well um I obviously at some point we'll probably offer a service that'll you know do some kind of benchmarking to Ure you know kind of like you you do when you look at like the AWS pricing tool you know what does my application need um versus what I think it needs okay and and one of the things about blocklist that I think is interesting is and you know this this is also on the website is if you're running the browser extension at

least um your your your expectation is that your computer is not going to be uh that the resources are not going to be used when you need your computer like uh you know if you're if you're running professional server infrastructure and you know the goal is just to provide uh compute resources to block list you're not concerned by that but but for a retail user you are so then I presume that applications are distributed across multiple nodes on the network so that when let's say I'm editing like a big Work video workload on my computer block this the block this uh ition is not running or at least is not using up my resources uh and and that compute is

demand is going somewhere else yeah so we use like a a similar kind of almost a reverse bidding mechanism that's you know similar to how a c providers bid on a deployment uh your node will advertise what's available to it like I only have 256 Megs of RAM available to me at this time can I take the work uh and then you know if we decide during the annealing process that nope 256 Megs is either going to cause too much turn or it's not going to complete fast enough you'll be moved out and we'll actually use another another node to do that work and you could very well be doing work at this point in time it's just that your Lego blocks don't have enough openings to snap in another app and this happens on

the Fly Like sort of Yep this so that's that's what you call Process uh un kneeling yeah yeah so we do a a dynamic uh filling process and then part of that is an analing ass simulation uh across all the nodes um there's a a big set of math that we do for Distributing over all of the nodes and finding the nodes that are you know more important to us at the point in time for that uh workload and it's all executed on the fly during execution requests so no two execution requests could be the same this is fascinating uh I think it's so cool and uh and I don't know if if you guys answer this but you know I've got some servers here I've got like a an little like Nuke computer here and I've

got a couple of raspberry pies running different things and Seb are you running a computer shop no but I mean like I've got like home assistant running on raspberry raspberry pie and I you know I just bought like a $200 um kind of little little mini computer that I'm using to run some Docker containers and and stuff like that you know can can I run a binary that will allow me to to utilize the resources on that uh on those computers because they don't have a browser right so I can't run a browser extension yeah yeah we do have a binary um it's just not incentivized uh right now the the benefit of incentivizing the the retail users is really us being able to get all of the use data you know working on on reducing Cil attacks as an example and

looking at how movements uh work in that but we do have full plans to release a binary I've also kind of a a sick thought is you know what happens if you deploy a Docker container that you're you're not utilizing 100% of and you put this agent in there and if you're running your website in the docker container then maybe 95% of the time you're you're not using the container we could fill that with blockless jobs or bless jobs excuse me yeah yeah yeah and I mean I could also run my website on on on on bless uh which which I was going to do on This Server so um yeah we'll have to talk about that that's that's interesting I'd love to dog through this and and and try it out for some kind of use cases like that

uh so one thing that happened recently which which kind of took me by surprise but what I think is super interesting is that you guys went you guys s of went from building blockless as an AWS to now building bless as as an application on salana uh tell me about that journey and um and why they moved to salana so I wouldn't say we are an application on salana um we provide the um compute infrastructure uh and because um the uh the execution it basically happens off chain and um it's done verifiably and for making this entire

system Fair um we made the Strategic um decision to move the um payment onchain and I think this is like typical for um some crypto projects you know to have the payment on chain the settlement on on chain um while their Tech infrastructure might be providing um you know the support functionalities Beyond uh the blockchain scope so that's exactly um the case for us and the choice of um moving the payment on salana uh is mainly because um it is currently the home of uh a lot of consumer apps and uh deeping projects

and in terms of the onboarding the user experience that's very smooth for especially people that didn't necessarily onboard to the crypto space through a metam mask wallet uh they first got their soul um say through moonay on their credit card and then they immediately got an account and then became part of this um you know economy and for us we see the potential for onboarding everyday users as easy as possible and I think that's the biggest benefit that we can provide for uh the industry is to make every uh user um participant of this shared economy yeah I'll just throw in you know

I think from oh from the technical perspective one of our concerns has always been reducing cost uh for participation and so when we look at large ecosystems like ethereum you know we do see that it's prohibitively expensive to get retail customers in the door they they just don't want to spend even 20 or $50 you know to onboard themselves so we've we've built multiple Cosmos chains and you know really what it came down to is if we're looking at a cosmos chain then it it also competes with our narrative on the other side because a cosmos chain still only runs really well with like 100 to validators you know and it's a very concentrated effort so salana seemed like a natural fit for us where we could do chain activities at a much lower cost and

we're less concerned with being tied directly to the ecosystem you know in lock step right so so salana handles the the the payment so I guess if if I need compute resources I'll make my my payment in Soul or or like I presume at some point the bless token on Solana uh payouts for users providing compute resources would be paid out on salana is that the only thing that's happening on Sal is just that that payment or are there is there other logic that's that's happening on the Chain um our whole registration orchestration in registration mechanism is on there so while while the actual execution and

picking some of the orchestration is totally offchain on salana we will do node registration node ownership we do Envision that there will be a point in module ownership where you may want to sell uh these modules to other people we have not really discussed in depth but we've already finished a lot of the the background technology to allow these wasm functions to interupt with each other and actual connect these blocks so we envision at some point there will be a Marketplace to sell these and all of that would also take place on salana cool so what are the use cases that this enables like what are people using bless for right now that's

exciting um well we've seen people are using uh for are say for example uh applications that requires uh low latency um and um especially Community powered infrastructure uh for you know running things that um is um not giving up the sovereignty and also it's verifiable uh for example we have clients using um our stack to build the um inference Network and we've also got clients building um an open AI Marketplace libraries leveraging hour uh Network stack um some other um use cases and uh I personally um quite

excited about is in the entertainment industry the uh gamings You' have um 10 games that are leveraging our stack to build the completely decentralized uh front end rendering um and the whole game experience so the all of them I think are uh going to scale exponentially in the upcoming a year or two and what what kinds of applications you know do you foresee in a year and as blockless grows and as the amount of Compu resources grow I mean are we thinking mostly web 3 types of applications where web 3 apps are um delegating some of the compute that they

need to block list so you know gaming I think is one uh use case or you know if we look at the way that aosh has uh kind of um you know progressed and and evolved it went from from being this kind of decentralized compute Network to now um really serving one primary use case which is um a Ai and and uh delivering GPU resources now this is a very kind of onweb 3 type of uh type of type of use case do you think that blocklist also be used by just like regular application developers that are looking for a cheaper and you know perhaps more decentralized or sensorship resistant type of infrastructure for

their applications yeah absolutely um I think the key benefit that our decentralized um Ash Computing Network provide is um for low latency real-time processing those kind of needs um like some kind of evidence we've seen that the thing on T Market is growing not just for you know the crypto is um with the immersion of AI um like projects um integrating AI or use some sort of AI um functionalities um this Bas has grown over 40% over the past year um and it's continued to grow at accelerating speed and all those call

for solutions that U require some sort of you know act processing um and also with the user generated content being a big part of it right um people um are more interested in opting into a community-owned um you know um infrastructure or community-owned um kind of property uh to be um also part of this entire product offering and we've seen um also you know problems with um arvr games these days they're not very popular right I mean I don't even remember it must be like a year ago when I play some sort of AR games um is because the load latency requirement

right now uh cannot be fulfilled with our existing um a Computing U infrastructure and it's definitely one area we see with decentralized compute Network it can fulfill um have a much better coverage um than any single centralized providers can possibly achieve and that is done also um with extremely high efficiency so lower cost right and also uh with those um growing need for the application not just like talking about whether they're web 3 or web two but ultimately they're kind of merging for providing better user experience and that's the future that we

want to be powering um and allowing um like users like me and yourself um and Derek to be part of I would say that it still kind of aligns with per demand usage envisionment um you know especially as AI starts to move off of gpus uh and back into CPUs um we don't want to really restrict ourselves to the actual underlying technology uh we prefer that where the most Supply is we can bring the demand and reduce the cost um because even traditionally if you use gcp today you have to reserve you know like an instance of the GPU or the CPU to sit on do your work and then disconnect and let it go to stop your

billing but we could automate all that based on the workloads uh that are coming in you said something that I'd like to double click on you said as AI moves away from gpus back to CPUs uh this is uh something I was not aware about what's Happening Here Yeah so there's a huge or is this work there's a huge movement now especially in the C++ World um to ensure that we're using what is already out there so a lot of the AI algorithm are being Rewritten to actually use new libraries that have been invented for the CPU um we're seeing uh formats like ggm uh which are specifically targeting low powered

devices like CPUs uh to ensure that inference can run really fast uh on them and so while gpus may be having the Heyday today um there is a large movement ensuring that we have all of this Legacy hardware and that can use it uh especially for AI models so there will be a time where both gpus and CPUs will be used tangental to do AI right now we're just kind of all sitting on on gpus and huge memory um but a lot of this work is has really been done before Innovation it's weird to say there was a lot of innovation in AI but the under underpinnings of it weren't very Innovative and now people are

coming back through and redoing a lot of what AI was built on um so that it is ready for other other Hardware that's been out there um and and one of those is is putting it back into CPUs interesting I I was not aware of this and so is the idea here that I mean like so of low uh demand inference would could get S done by by CPUs and then the higher demand stuff would get done by the gpus right because you don't need to run an llm on on a GPU it's it's just a text Matrix and a text prediction and if you can get something like 50 tokens a second out of that then it feels like natural speech between a human and and

the computer but there are other things like massive renderings and composits of scenes and so right now we've all been looking at 512 x 512 or one megapixel pictures and we've seen you know stable diffusion 3 come out and do you know four or five megapixel pictures but really where gpus need to go is is rendering spatial 3D using Ai and I think you'll start to see the separation where we'll use signal processors dsps audio processors to do simple things like text inference object identification um you know something simple like providing you a recipe because it can combine multiple recipes and then real life Real Time 3D start to

move towards gpus um there's also a technology that we don't ever really talk about in line with Ai and that's called gloss and splatting it's a a new 3D technology for video games um but it requires large sums of memory like 64 128 gigs of memory and so once this makes it into consumer Hardware then I think we'll see the AI rendering explosion happen um you know in parallel with that so it sounds like in the AI space there's this push for optimization of the resource um and the like the the resource Supply and the actual demand of the computation and what it sounds like is

that currently there's there's too much compute being used for things that could utilize simpler uh less powerful compute yeah I think that's a a perfect way to say it we just have a glut of Supply um that could be totally useful for many of these use cases interesting okay um I I'd love to get you and um and Greg from AOS on a podcast once and oh yeah yeah that would be fun this kind of stuff yeah yeah we just did a podcast with him at uh I did one with him at cosmoverse and we didn't talk about iOS at all we talked only about like compute and like building nuclear

py plants it was a really fascinating conversation um but uh I learned a lot so you know coming back to the extension what's running behind the hood here like what is running in my browser uh that is allowing me to participate and run these computations yeah so we're you know the magic is that uh that Google has done a lot of work on Chrome and the Chrome base and so um as of chrome 126 we now have what's called Web GPU uh prior to this we had to draw a bunch of stuff on the thing called canvas and use you know a GL interface between that but now we

have raw access to the GPU directly in the browser um and then so you know we use uh the nice thing about the extension is it allows us to live outside of the of the browser life cycle so you don't have to keep your tab open things like that we're able to to run in the background um JavaScript has also come a long way uh and so we run an entire P2P client uh in that browser um and then it executes wasm in JavaScript yeah PTP client uh in JavaScript yep that's correct yeah and that connects to our goang uh back end uh using the same P2P protocol um and then we utilize the fact that the browser uh the V8 engine itself has its own wasm uh virtual

machine and so we've written uh what would be an analogy to our our full binary VM uh in JavaScript shimming around their V 8 engine so we we do have basically a full uh and you mentioned this before b7s so that's blockless shortened it's a play on kubernetes katees um we do have that entire node then shoved into that extension um operating just exactly like it would if you were running a native node it's obviously trimmed because it's in the web browser and there's some things we're not allowed to do in the web browser um but it can execute about 80% of of the payloads that we've you know come up with so far right so so this b7s block blockless

networking demon what is what is that do exactly like I mean I see I can I can install that as a Docker yeah why would so you can think about it exactly like a a blockchain node with no state so okay it runs in the background it connects to a peer-to-peer network uh and it listens for messages that are coming off of this network to to do some things with uh one of them is dynamic consensus formations so we do have like a simpler version of tenderman we have uh a similar thing to to chain link functions with an aggregation mechanic and so the nodes will then you know shake hands with each other become friends move in through a quorum execute

and then return those payloads all together um without ever keeping sted okay interesting and and yeah sor just got a tsunami warning and all my devices are ringing a tsunami warning sorry we'll wrap up in like five minutes but like you should figure out what the going on so you don't get yeah scared the out of me yeah it's like one of these Amber Alerts um when you think you're still in but you're literally in San Francisco okay um yeah so SDK I also want to talk about the SD a little bit

and how developers can use it and you know the supported languages and basically like this is the this is the gateway to block list this is how I keep saying blockless bless uh and how people build applications using uh using well on on blast right yeah not not just you I'm I'm having a hard time rebranding myself I mean your socks still say blockless yeah reading it's a it's a move over uh for sure yeah um yeah so our you know so we we support wasm at the center of this so any language that can compile to WM a developer can use that could that could be C++ python rust go uh assembly script um you know the list really goes on net

yep um and then we provide you know like a sample uh API that they can interact with um but but honestly what we found I think over the last couple years especially being in this space is that rust and goang Engineers are very Niche and there's you know a small concentration of them that even want to build kind of daps or applications you know that are consumer facing and so the last uh probably three to three to six months we've been working really hard on bringing JavaScript as a first class citizen over to the bless Network so you can develop today using your your typical typescript or JavaScript uh and build that we pack package a jvm a JavaScript VM with your assembly

and so no matter where this portable executable WM is is executing it always has the correct version of JavaScript that you've written your application against um and it's it's pretty small it's like a like a you know 25 megabyte engine that ships with the rest of your um assembly um and and so a developer's experience today is really downloading our our BLS uh command line uh installing that it's very very quick you just ini initialize a function you can choose the JavaScript template we have a several different templates for you to choose from um and then you can deploy execute all locally and then uh when you're ready publish it uh to our ipfs Gateway and then send execution requests

uh to the nodes so I mean I I could build a website I could build a uh like a nextjs app and deploy it to plus yeah absolutely we have support for uh Vite uh nextjs uh several different front-end Frameworks that are built into that as well um and then we have Dynamic routing available as well so it's not just a static website you can do you know user interactions as well uh like you were used typical server client interaction right and but do I get an endpoint for this where I can like call it up in a browser yeah you do so today um you know like like all big web 3

projects we use cloudflare they're they're really great obviously for for a number of reasons um but you could bring your own domain or we will assign you a common name um typically starts with like your function name a hash and then it it'll sit at like bs. um we run uh a portable web 2 Ingress so what's happening is that we have a web two facing Ingress that you could launch yourself you don't have to rely on ours um and then it connects to the P2P Network and then everything from the Ingress back is all web3 P2P uh getting the execution asking for the results uh and then it's transformed into you know web friendly uh stuff uh in this Ingress and then shoved out to

the consumer the Ingress could be anything it just web 2 is like the prime example use case right now but it could be like a native app on your Xbox uh interfacing with P2P and then showing some kind of Xbox experience very cool so what's what's coming up like what's on the road map what can people expect in the next couple of months oh one one thing I wanted I I wanted to ask uh before we get to that what happened to the button yeah so for some contacts we run a campaign before the release of our Chrome extension node uh it's a web browser a web page node uh it's a demo

game we call it the button as users are on the web page as they're playing the game they're automatically contributing their compute to the network and that game the button uh depending on when and how you click um and that will um basically convert to different uh consequences um that uh you would see later um so long story short uh you guys a whole month ago this was like a campaign that was yes run yeah this is a campaign it's a demo um game on web app that just showcase you know what it means for people to automatically

contribute compute uh without even them knowing that simply by using an application yeah svision users yeah it may have been hidden to most users but if you were looking at the map campaign uh your node was actually providing compute for other players yeah okay yeah and this is like really what we envision the future is for um applications um we call it the embedded node and what it means is for the future applications to be really uh user powerered and there's really no overhead in you know say if I'm a node operator

or what um you're just users of the application and you're automatically powering and hosting the next user that uh comes online to use the application um and I think this is really the ultimate the end game right probably like um when you asked about the road map that's like the really the end game where we envision the new internet uh should be and that's coming up um in um the next few years awesome and so yeah let's talk ni let's talk about the road map what uh what could people expect from blocklist in the next couple of months and when when

token uh when all the great all the good stuff yeah so uh we are launching the mainnet um in q1 2025 so it's going to be a great year for us and then scaling from there uh we will um on board more uh users on the supply side by introducing the uh desktop client uh the mobile client um and even consider uh working with some of the hardware uh device providers um and on the demand end uh we will launch together with um a couple of those uh early co-build partners and then scale further with um

some off-the-shelf um templates and packages for some common use cases um like the ones that we've um built for generic inference Network running AI agents um and some for the game um kind of toolkits that developers would typically need and then further smoothing out the uh user experience on the developer end um through the CLI node um and um dere could probably share more on the technical side what um additional support with that we are uh providing in terms of multil language and um environment um and all those good stuff yeah especially from the technical side in the next few months you know we really want to start continuing to

increase our reach and developer experience uh developer onboarding ensuring that it's easier to get into our ecosystem so there's a lot of focus on that uh as I mentioned we have uh some plans to release modular wasm files that plug into our ecosystem as well so people can start if it if a d app isn't something that you know is in your grand grander ideas maybe some kind of small function that you could profit off of while allowing other people to build uh scalable D apps is and and really starting to empower uh all facets of that Creator Community amazing well guys uh thanks so much for coming on and providing this really exciting update on

bless uh I'm finally going to get it I'm going to start saying bless and um yeah happy to have you guys on in a couple months when all this uh when when mainnet uh goes live and more applications are using the network and uh please yeah keep us updated awesome yeah let's stay blessed yep stay blessed absolutely cheers thank you s [Music] [Music]