Takatoshi Shibayama: 0:00

Hey, hey, hey, welcome to this week's episode of Blockhead's Blockcast. I'm your host, Takatoshi Shibayama. I'm also the head of APAC for Ledger. I aim to uncover the creative, intelligent, and radical minds who are shaping the crypto industry today. I'm as crypto curious as anybody that's tuning into this show. We're doing this together, guys. Let's go. Mr. Abhinav Ramesh and Mr. Vinay Mohan, welcome to the show.

Abhinav Ramesh: 0:31

Thank you so much, Taka. Great to be here. Thank you, Taka-san.

Takatoshi Shibayama: 0:34

Yeah, you're respectively a CEO and a COO of the company. I would really love to understand your business because you're working on things all about AI agent marketplace and also a deep in marketplace. It's about, you know, marketplace being a marketplace for GPU compute. You know, before we dive into that, which is all very exciting, I'd love to understand your journey into this space, how you got into crypto and then so how's your journey been so far? So let's start with you,

Abhinav Ramesh: 1:01

Sure. The journey into crypto started quite early, started around 2017-2018, where we built many in-house... custom products on crypto. Some of the highlights were we built a very large social networking platform called Murmur, which was at that time on the EOS blockchain. It had over 2 million transactions performed on the network, and we were at one point one of the top dApps on the EOS blockchain. There was a kind of pause on the blockchain for about quite a few months. Post that, worked on quite a few enterprise-based use cases, worked on tokenization projects for the gold industry in carbon credits, and after having Having discussed with Vinay and with the fast growth of AI and being into blockchain, wanted to see where this intersection was around AI and blockchain. And that's where we found, you know, there is a lack of adoption, lack of access, lack of ease of use for AI users. applications on blockchain and hence we came up with and started Node Foundry.

Takatoshi Shibayama: 2:01

Nice. And Vinay?

Vinay Mohan: 2:01

A

Vinay Mohan: 2:02

So my background is about 15 years in management consulting before I found myself in the world of Web3. I was in Singapore in 2016 when I became the first employee of ConsenSys in Asia and ConsenSys was a very small organization at that time and my mandate was to hold the flag for what then soon became known as Enterprise Ethereum. So because I have a background working with businesses and governments, I was brought in to sort of help build a solutions and services business aimed at how we can take the blockchain away from just crypto and try and apply to different use cases in enterprise and government settings. And Singapore was the perfect sort of petri dish to make that happen. So over the next four years, I participated in a number of, you know, world's first projects. Significantly, the monetary authority of Singapore was working on central bank digital currencies, was trying to use blockchains to find out how a central bank could in some terms be a decentralized institution, which of course has then become lower in the world of payments technologies. I also did some work with the media regulator, with the Sovereign Wealth Fund, built the world's first security token exchange to use the Ethereum mainnet as a tether at a time when the protocol wars were happening. So I sort of bore witness to the rise of the enterprise side of blockchain and then moved on to specializing in central bank digital currencies. I helped launch the world's CBDC in production, the Bahamas Sand Dollar. And since then, I've been looking at ways to bring in enterprise and government clients into the very wild world of Web3. That's sort of my interest right now.

Takatoshi Shibayama: 3:38

Yeah, it's a wonderful journey. And right now, obviously, AI is a huge buzzword. It's very much talked about outside of Web3 as well. Now, coming into the crypto industry, we do hear a lot of mixtures of AI and crypto quite often, especially past 12 months. What do you think is really the use case for AI and crypto? Because obviously there have been some AI bot that actually created its own meme coin and launched it, et cetera. But that's just more of a hyped up story. In my mind, it's not really a big use case in my view. But what do you guys think that AI and crypto can really solve real world problems?

Abhinav Ramesh: 4:19

I think one of the fundamental use cases would be providing cheaper, easier access to AI infrastructure through the use of blockchain. And that's where DePin comes in, where multiple networks are available today through which you can access far cheaper compute, storage, bandwidth, decentralized GPUs. than what is provided by centralized parties in addition to that i think one of the use cases would be privacy so if i want life large enterprises are to enter the decentralized AI ecosystem there is going to be a great need for privacy and simplicity and ease of use so if i was an enterprise i would want my data to be private and if two competing enterprises are on decentralized infrastructures, decentralized infrastructure networks. I would want the data of my enterprise to be private and not visible to the other. So I think privacy and access would be two large use cases that I see.

Takatoshi Shibayama: 5:25

But that's only kind of more of the data storage side of AI, but the intersection between AI and crypto is well debated these days. I can see like AI agents working on behalf of me, trying to do some workflows, payments, et cetera, on my behalf. Maybe that's kind of where crypto comes in. But Vinay, what do you think is the crossroads between AI and crypto?

Vinay Mohan: 5:47

So I think the other sort of first principles discussion we should have here is let's go back in time and think about why Web3 was what it was. What was the narrative behind decentralization? We all sort of experienced the ideals of a shared economy. We came from the platform economy to sort of a more distributed, you know, silo-centric economy. And then we talked about the promise of the shared economy. I think we're far, far away from unlocking the promise of that. And that's sort of what's motivating us to do what we're doing with Node Foundry. AI presents one more opportunity on top of what we already have discovered with blockchains to democratize access, not to the same 3.5 to 5% of the world's population that's using it right now. We're talking about the bottom of the pyramid over here. We're talking about people who've never been able to access this type of technology and essentially normalize being able to use that.

Vinay Mohan: 6:38

So that on its own, I think, merits some discussion because now we're talking about an exponential growth in scale. It's not the same sort of dog-footing environment where one is using Web3 startup tools, tooling and technologies. We're not talking about people who don't necessarily feel like they belong in the room. We're talking about the Web2 audiences. And many of them are extremely sensitive to pricing, to the terms offered out there. Not all of them are very happy. And I think we can get into a little bit more about why compute in particular is getting very expensive and how it's getting harder for the bottom of the pyramid users to get into this space.

Vinay Mohan: 7:12

So in a world where optionality is important, in a world where price sensitivity is important, we need more variety. We need ideal sort of solutions of the convergence of AI and deepened infrastructure. We need to be able to produce solutions out of this distributed economy and essentially help people access compute, storage, video, agents, LLMs, whatever be it. So that's sort of the world we're looking at right now. We take that first principles approach because the distinct difference we want to make with Node Foundry is not to be able to just sell to the people already in the room, but to bring in a far greater number of people who don't necessarily fit in or have no pathways to get into the Web3 space. That's the target we want to go after.

Takatoshi Shibayama: 7:56

I totally agree with you with the topic of equalizing the playing field. Right now, I think most of the crypto industry are building things for the Web3 and people It's not really thinking about the Web2 or anybody outside of it. A lot of the platforms are very complicated, lots of jargons, lots of abbreviations that nobody really understands outside of the industry, especially even the word deep in is very foreign from the outside world. Back in 2017, there was projects that were building decentralized data storages. That's what it was called back then. And you have your likes of Filecoin and Bluzelle and all these companies building it. It never really took off in a way that we actually really thought. Nobody was actually building their platforms on these so-called decentralized data source storages. Nobody was using it. Back in those times, I was thinking, well, if people in Web3 are not actually using the products that the Web3 industry is building, so what's the point of all this, right? So what do you think didn't work out back then that you think that could work out today?

Vinay Mohan: 9:01

Maybe just a couple of features. I remember back in the day when we used to speak to... government audiences about blockchain, Web3, crypto, and sort of try to educate them and tell them about this new world that's emerging. And this question was being asked quite often. This is all so good and so true. Why hasn't the world still taken notice? I recall a time when, in the early days, when Metamask hadn't been, you know, Metamask Mobile, and we were still using a browser-based plugin. Two features which everyone would sort of agree was sort of lacking solely in the crypto ecosystem at that time, which since then has been, to the ecosystem's credit, been developing. One was just UX, right? For the want of a better word, just being able to access things in a more visually friendly way, right? And not having to go through this convoluted workflow of multiple clicks and installations and in a very code-heavy sort of work. So just the UX for the broadest possible definition of that was missing. And the UX that is desired by people who are not in the room is vastly different. We have to greatly simplify a lot of this for the larger audiences.

Vinay Mohan: 10:05

The other was just education. There was too much technical jargon. There was too much crypto bro-speak happening. And it's incredibly hard for people who are genuinely trying to discover how to work something like Deepin. The physicality of what DePin has to offer is a great leveler. Everybody understands that the last mile use case for computer storage or video rendering or an AI model, that's all understandable. But the pipes that take you there, all of the little alleyways that people have to get into, that's not a very educative process. And these were two big factors that hindered both the crypto ecosystem from growing, but also now that we see in deep. And so the word I think here is abstraction. And that's what we're trying to do. We're trying to take away the complexity of all these different terms. I mean, everyone who's been part of Web3 has said this sometime in their life, right? We don't want the word blockchain to be used. When you get on the browser right now, you don't say, hey, I'm using TCP/IP. It's normalized. It's something that's sort of folded away and faded away into the background. We all wanted the same with blockchain. Unfortunately, I think the opposite has happened. And we certainly want to get to a world where people aren't concerned about the chain or the token, the smart contract or the wallet. What they are concerned about is the use case in its purest form and everything else is abstracted away. That's what we want to try and do with our offering.

Abhinav Ramesh: 11:27

In addition, I would say the timing as well. With the rapid growth of AI in the last two years, there has been greater demand for H100s, GPUs, compute power. And given the high costs of setting up data centers, there is given free network space available. I think that has coincided a lot with the fast growth of DePin, where you can rent out compute, storage, bandwidth, data marketplaces, maps marketplaces, rendering marketplaces through different deep end networks and lower your cost as well, lower the time in which you can perform your action. So I think the timing also has something to do with adoption.

Takatoshi Shibayama: 12:11

Absolutely. And if I were building an AI agent today, obviously the data storage will be on your AWS or Azure, Google Cloud, etc. Those are very well tested, very secure at the moment but it's not everywhere. And it does come with a cost. So if I were to use your traditional AWS's of the world versus using Deepin, how would they differ in terms of like price, access, and also coverage?

Abhinav Ramesh: 12:38

So as of today, DePin is growing at a really tremendous pace. So there are over 150 projects currently active on DePin. Out of that, about 40 of them provide decentralized compute systems. And Quite a few of them also provide decentralized GPUs or GPU networks through which I can use these D-Pins. So I can potentially access all the pillars needed, which is storage, bandwidth, compute, just through DePin, but rather than a centralized system like you mentioned, like AWS. But the benefit of DePin would be that it's much cheaper.

Abhinav Ramesh: 13:15

You pay as you go and the disadvantage would be that you might have to use multiple networks so if a particular network is clogged if a particular network does not provide enough storage or if it's not to your liking and you would like another network you would have to use multiple networks in order to get your tasks done whether it is running a model or whether it is you know storing certain data or basically having kind of a decentralized cloud infrastructure for your company or for yourself as an app developer so I feel the disadvantage would be I would have to use multiple networks, in which case I would need multiple wallets. I would need multiple tokens. I would need to learn multiple smart contracting languages. And that adds a lot of complexity. I think centralized systems do not have that kind of complexity. But decentralized systems do have a great cost benefit. And with upcoming privacy solutions such as ours, there would be the benefit of privacy as well, which is already there in the centralized systems.

Takatoshi Shibayama: 14:11

Yeah. And, you know, you talked about chain abstraction. Obviously, they're all crypto projects, right? So they come with tokens, tokenomics, etc. You know, for the Web2 industry, that is not really relevant. They don't really care about tokens. They don't care about chains. And abstracting those things are very important for the adoption of the people outside of the Web3 industry. How does Node Foundry's architecture work to abstract all those complexities away?

Vinay Mohan: 14:38

Sure. So we've taken a layered approach to the architecture. The speak. We have a beta out there, which is usable by anyone who visits nodefoundry.ai. Please take a look when you can. But to keep it simple, we are treating Node Foundry as a two-sided marketplace. So any user, let's say the actors could be developer or a startup or a small and medium enterprise, any of them could basically come and discover various services using Node Foundry. So Node Foundry becomes a single point of access. It also aggregates multiple use cases, multiple deep in and crypto AI use case is in one place.

Vinay Mohan: 15:14

So to add to what Abhinav said earlier, rather than you as a user individually going and working with a separate compute platform, a separate video rendering platform, a separate storage platform, a separate agent platform, you would pretty much have a consolidated menu of all of these applications at your fingertips with Node Foundry. So the aggregation aspect is taken care of. The idea is to provide a single API clickable interface for users so that they don't have to go and do too much of back-end coding, any sort of heavy installation at their end. It should really be as simple as a single-click drag-and-drop as far as API abstraction is concerned. So the third part is abstraction, essentially, which is the user is just looking at the Node Foundry solution. And if you go to our beta today, you'll see there's a menu of services. You can go to Akash. You can select any one of the 10, 15 templates from Akash and literally have that instantiated through a single-click API key. So we're trying to make it as Web2 as possible in that sense.

Vinay Mohan: 16:11

And in this process that I've just discussed, described right now, at no point is the user actually dealing with any chain. They're not actually using their wallet to pay with any tokens. They can securely access all of these services with credit or debit cards. Node Foundry would do the tokenomics and the exchange and the backend on behalf of them. So the user's not really seeing any blockchain at this stage. They're not really being bogged down by the detail or the technicality. They're not feeling left out. They are seeing this as essentially what feels like a browser-based cloud-hosted solution, and they're able to get what they want through this process. So we do all of these things, but we also enable privacy preservation.

Vinay Mohan: 16:51

So the idea behind our architecture is to use solutions like 0.0 proofs. So if you're an enterprise, for example, if you're a company and you have to host financial data across decentralized infrastructure. We can take care of the trust problem, right? So we can help you tether your financial data with a considerable degree of privacy. We can make sure that we manage the orchestration between all of the different solutions without your data actually being used or capitalized by any of these third party players. And on the inside, we have the option of also settling with our various stakeholders using, you know, so we can buy tokens on their behalf. We can use their tokens to do the internal settlement. We also have plans to have a token that becomes a system of settlement between us and our partners. So that's just a quick sneak peek under the hood. And this is how we plan to make it simple for everyone to access our fund.

Takatoshi Shibayama: 17:42

And how does the revenue model work for you guys? So obviously, I can assume that you'll be charging these marketplace for AI agents, you'll be charging them for the compute. Do you guys make money off of the tokenomics that are in the background as well?

Abhinav Ramesh: 17:56

Like Vinay mentioned, we are an aggregator, right? So we provide access to deep end solutions through simplified APIs. So you can access multiple deep end solutions through our APIs. And we simplify that access through aggregating all these deep end solutions or multiple deep end solutions across different chains. When there are calls to APIs when there are uses, when these Deepin solutions are used through our platform. That's where we charge a transaction fee. And part of those transaction fees goes back to our token holders.

Abhinav Ramesh: 18:26

In addition, our token holders can also stake their tokens and receive these transaction fees as transactions keep happening through our APIs. And second is we also provide, like Vinay had mentioned, abstraction. So you can think of it as the three A's. So access, aggregation of DePin, and abstraction to simplify the use. And then the fourth layer would be privacy so like when i mentioned there is difficulty or there is hesitation to put data or sensitive data on chain and zero knowledge proof based privacy will help you build that trust and you will be able to put your data up on chain and we charge not only the transaction fee on the buyers or on the user side so from the deepen solutions that we aggregate we charge a transaction fee because we help increase usage and on the user fee as well we charge a much smaller transaction fee because we are helping the user access deep in networks through a much simplified user interface

Takatoshi Shibayama: 19:19

Makes sense. And then you know if i'm a Web 2 builder and I've come to your platform, could you take us through the whole journey of like how it actually works so I have I build an AI agent program I'm ready to kind of deploy it I come to your website how do I actually use it?

Abhinav Ramesh: 19:36

So let's take that workflow. So Ii have an idea for an ai agent i would like to build a particular ai agent for example i'm a small medium business or I'm an individual developer. I come to Node Foundry, I see I can get compute from XYZ network, I can get storage from a particular network, I can get decentralized GPUs from a particular network. So what I do is I call these three particular services from Node Foundry. We in the backend take care of optimizing which network you can call for, which has the least bandwidth, which is most cost efficient for you at that point of time.

Abhinav Ramesh: 20:12

Once you call, not least bandwidth, the maximum bandwidth and the So once you call these services, say you want to deploy a particular model for your agent. So you can choose from the list of models available or you can go to platforms. There are multiple networks like BitTensor, which offer machine learning models as well. So you can pick any of those models. And once you go to deploy, we have that orchestration layer where we orchestrate what you can aggregate in terms of compute, in terms of storage, in terms of bandwidth, in terms of GPUs and which model you'd like to develop. And then you provide the data, you train the model, and then you will be able to deploy it on a decentralized cloud. But we simplify this entire process so that you do not have to deal with four or five different tokens or four or five different types of chains. And you have a single place where you can access kind of decentralized AI or decentralized AI infrastructure through a much simplified mechanism.

Takatoshi Shibayama: 21:08

And how do you think the future holds for decentralized AI? So obviously, there are going to be lots of centralized AI. How would the decentralized AI landscape look different from the centralized version of it?

Vinay Mohan: 21:21

Hopefully with sufficient abstraction and efforts that Node Foundry and others are putting in, users will not see the difference. I think that's our ultimate hope, right? That we don't see this line between centralized and decentralized anything. We give users the full knowledge of what they're getting into. We give them safety. We let them know what's happening with their data, and we give them high-performance results at the end of the day. That's essentially the simple sort of destination that we want to get to, as do a number of others.

Vinay Mohan: 21:49

It's worth thinking about this from the perspective, again, and this is where we have to sort of open our eyes towards, you know, if we see this from the view of the existing Web3 community and the people who are in the room right now, I think it's going to be highly limited. We are a a confluence of a number of competing solutions. It's going to get messy and exciting at the same time. But then if we think about this as, how do we go after millions of people in emerging markets? How do we go after people who haven't had a chance to use this and don't necessarily have the effort or the capital to go and work with a large company or an AWS or a Microsoft, even though those centralized solutions will be going at these pretty hard?

Vinay Mohan: 22:29

I think we have a chance to help buyers and sellers who would have otherwise never met get together and do something interesting I think there's a whole range of economic possibilities out there that no one has uncovered before and that's why this is exciting for all of us because just like we would say back in the days of security tokenization, right? We're trying to unlock frozen capital. We're trying to unlock frozen value. We're doing the same thing here. We're trying to unlock frozen commerce between two parties who would otherwise never speak to each other. So the future, I think, should be blurred. We don't need to know what's happening behind the curtain. And ultimately, users should be able to see several options in front of them and extremely simple ways to click and access and use the actual service.

Takatoshi Shibayama: 23:12

In early tractions, what types of AI agents are you starting to see on your platform or even outside of your platform? What are some of the exciting stuff that you guys actually see?

Abhinav Ramesh: 23:22

A lot of interest in, of course, agents today. Over 10,000 plus agents have been created just, you know, and many of those agents have their own tokens as well, like you had initially mentioned. There is a lot of interest in creating different agents with crypto-specific use cases, of course. So DAI or DeFi plus AI automated trading solutions automated staking liquid staking solutions those are some kind of real key points that we see because a lot of the data that you need for these agents is available on chain it's quite transparent so you can use multiple providers that give you these APIs that from where you can collect a lot of the data to train these models or to train these agents that is somewhere where we are seeing a lot of interest.

Abhinav Ramesh: 24:04

The second is we are also seeing interest on non-crypto side but specifically those wanting to create ai agents but using some some data from the blockchain or some data from deepened solutions. So deepened data marketplaces that provide video data, maps data, weather data, analytics or draining AI models on that particular on that data, on those types of data sources and providing insights is something else that we're seeing a lot of interest on.

Takatoshi Shibayama: 24:33

And do you see anything outside of the Web3 industry that are kind of looking at this? Have you had conversations with Web2 players to be using your platform? Is there an understanding of these decentralized data storage solutions?

Abhinav Ramesh: 24:47

We've had quite a few conversations with Web2 players but I think the key will be if we can simplify the Web3 journey enough for them. If we can like i said the four the three a's so provide easy access provide aggregation of the best deep in solutions in the most cost effective manner abstract away the complexity of web3 of chains of tokens for these users and provide the privacy layer if we can do all of this and we have a very strong back end through which we can orchestrate the highest the best uptime as well as the best kind of service for these web2 players i think once we are able to do that at a sufficient kind of confidence level, there is going to be much larger adoption from the Web2 players.

Takatoshi Shibayama: 25:32

Right. And we're kind of at the end of the episode. Are there any parting messages that you'd like to say?

Abhinav Ramesh: 25:38

We are building quite fast. I would say check out Node Foundry AI, which is our Twitter handle, which is where we'll have most of our updates. We've got a Telegram group as well with the same handle. You can join that and stay tuned for updates. We are hoping to really bring this to life or bring our vision to life to simplify Web3 deep in for a Web2 audience. Also, of course, increased adoption of Web3. I think if you can stay tuned to our socials, you will see updates and we're really excited to be building this.

Takatoshi Shibayama: 26:12

Cool. And Vinay, anything from you?

Vinay Mohan: 26:14

Yes, and just to add to that, if anyone's listening and wants to be a partner to Node Foundry, please reach out. We are actively looking for partners on both sides of the marketplace. We'd love to work with large enterprise audiences, managed services companies, government agencies who would like to delve into the world of Web3 crypto, AI, and DePin, but don't know how, we're very happy to walk you through that education process. And on the sales side, we are trying to work with a number of anchor players who have AI solutions or DePin solutions to market, but don't have the vertical GTM capabilities, and that's lacking today. So we would love to help you get to market faster to audiences that you never thought were possible to reach. And that's sort of our goal for the next quarter.

Takatoshi Shibayama: 26:58

Great. Thank you very much for your time. Thank you so much.

Vinay Mohan: 27:01

Thank you, Taka-san.

Takatoshi Shibayama: 27:02

And thanks for listening. If you like what you hear, give Blockcast a like and subscribe on your favorite podcast channels, Spotify, Apple, wherever you listen to them. And for all your juicy Web3 news, keep updated on blockhead.co. Catch you all in the next episode.