Latest / Crypto Coin Show / Decentralized AI Agents & 400,000+ Nodes: Gaianet’s Vision for Blockchain-Powered Intelligence
Transcript
- 0:00Speaker 1: I'm Ashton Addison from the Crypto Coin Show and today
- 0:03Speaker 1: on Blockchain Interviews with Shashank Shripatta, co founder and COO
- 0:07Speaker 1: of Gaya, here to talk about AI and blockchain, AI agents,
- 0:12Speaker 1: decentralized AI, and all the functionality that AI agents are
- 0:15Speaker 1: bringing into blockchain and just to the world in general. Shashank,
- 0:19Speaker 1: thank you so much for taking the time.
- 0:21Speaker 2: Yeah, thanks for having me, Ashton, very welcome.
- 0:24Speaker 1: There's been so much functionality increase that I've seen versus
- 0:30Speaker 1: the talk over the last year or even two years
- 0:33Speaker 1: of you know, AI agents emerging and actually driving functions
- 0:38Speaker 1: that can save us time and automate a lot of
- 0:41Speaker 1: things on the Internet. And I know your team is
- 0:44Speaker 1: at the forefront of innovating in AI agents. I'd love
- 0:48Speaker 1: to hear to start off a little bit about what
- 0:51Speaker 1: you and the team have built at Gaya, how it
- 0:53Speaker 1: relates to AI and AI agents, and then we can
- 0:55Speaker 1: dive into all those topics.
- 0:57Speaker 2: Yeah, sure, thanks for having us. Yeah, Guy, is I
- 1:01Speaker 2: think creating the conversation and pushing the conversation on agents,
- 1:06Speaker 2: what they can do, and why decentralization is the only
- 1:10Speaker 2: way forward for AI. So I can explain Guy a
- 1:14Speaker 2: bit guy is an open source DevOps framework. Originally, it's
- 1:19Speaker 2: about three lines of code helps you set up your
- 1:21Speaker 2: own agent nodes. We make it super simple, cheap and
- 1:25Speaker 2: easy to set up agents, or you can think of
- 1:27Speaker 2: as as the strike for setting up AI agent nodes.
- 1:31Speaker 2: So an agent is essentially a node. Agent in our
- 1:35Speaker 2: case is essentially the box where the house is the
- 1:41Speaker 2: open source l M, the vector DV and any hardware
- 1:47Speaker 2: set up underneath it. The secret sauce there is the
- 1:50Speaker 2: wassome edge run time, which are chief scientists to Michael
- 1:54Speaker 2: you want created. We do maintain that alongside the the
- 2:01Speaker 2: Linux Foundation, and essentially you can plug in any open
- 2:04Speaker 2: source LM. It acts as a virtual machine and it
- 2:07Speaker 2: finds any hardware you're running underneath it, so you can
- 2:11Speaker 2: run it off in aku I CPU all the way
- 2:14Speaker 2: up to in video or a WUS or a bit
- 2:17Speaker 2: tense or anything in between. So we make it super
- 2:20Speaker 2: cheap and easy to fine tune learning models to any hardware.
- 2:28Speaker 2: On the other side of the spectrum, I think the
- 2:30Speaker 2: market we're trying to go after is proprietari your data
- 2:33Speaker 2: sets of humanity. What that means is, you know, open
- 2:36Speaker 2: AI kicked off the general of learning race. We're grateful
- 2:39Speaker 2: for them, but we think that it's mostly script training data.
- 2:44Speaker 2: It's a kind of non specific data. It's not proprietary
- 2:48Speaker 2: data sets. So we think that most of the humanity's
- 2:53Speaker 2: knowledge is under the proprietary data sets that enterprises, government's individuals.
- 2:59Speaker 2: It's too hold and they by large, those entities have
- 3:04Speaker 2: not kind of hopped into genital learning models. We think
- 3:09Speaker 2: that's because in order to open them up you need
- 3:13Speaker 2: open source and you need web three because the data
- 3:16Speaker 2: it's way too valuable. Otherwise, these entities do not want
- 3:21Speaker 2: another monopolist like open AI or Google or anyone else
- 3:26Speaker 2: sitting between them and their most precious asset, which is
- 3:30Speaker 2: their data. So we're obviously an open source. But on
- 3:33Speaker 2: the one three side, we are an inferenced network. We
- 3:37Speaker 2: allow for the inferencing of agents on chain. We allow
- 3:41Speaker 2: for the attribution, validation, monetization of inference on chain. That's
- 3:45Speaker 2: our essentially our niche in the AI tech stack, and
- 3:51Speaker 2: we do that through a to main node architecture. Essentially,
- 3:55Speaker 2: for example, Berkeley University of Berkeley, where some of our
- 3:58Speaker 2: researchers conducted, has its own domain Berkeley AI, and it
- 4:06Speaker 2: acts as a supernode, a TCP IP router to write
- 4:10Speaker 2: any traffic or requests to Berkeley to the right mix
- 4:14Speaker 2: of nodes that are underneath it. So every professor at
- 4:18Speaker 2: Berkeley is operating its own AI agent node. So if
- 4:21Speaker 2: you have a chemistry question, your request gets routed to
- 4:24Speaker 2: the chemistry professor's agent and any additional computers bolstered both
- 4:29Speaker 2: internal and external to that domain. So in that way
- 4:32Speaker 2: we make it very efficient to match mL compute with
- 4:35Speaker 2: mL models, and people can also steak their tokens at
- 4:39Speaker 2: both the domain and node level to earn seeking rewards
- 4:44Speaker 2: for maintaining and validating. And there's also a slashing mechanism
- 4:48Speaker 2: we're developing essentially using eigenlayer for both those mechanisms. Igain
- 4:54Speaker 2: is actually a major partner of us. We've been building
- 4:57Speaker 2: on them for a little while. And yeah, I think
- 5:02Speaker 2: overall the architecture of Gaya makes it easy for you
- 5:07Speaker 2: to create an agent and then share and monetize that
- 5:10Speaker 2: agent on a Web three landscape.
- 5:13Speaker 1: Wow, it's a lot to cover. Shashank and I love
- 5:18Speaker 1: the little bit of insights into the enterprises and these
- 5:22Speaker 1: huge companies that have so much valuable data, but of
- 5:25Speaker 1: course they don't want that data to be in open
- 5:28Speaker 1: AI for competitors and everyone else to look at, And
- 5:32Speaker 1: how do we set that up for them to utilize
- 5:35Speaker 1: their data more efficiently and can have a competitive advantage
- 5:39Speaker 1: without giving it to everybody. So I think that's where
- 5:42Speaker 1: the decentralization or the permission part comes in. I'm curious
- 5:48Speaker 1: from a regular user perspective. Let's say, no, there's hundreds
- 5:52Speaker 1: of millions. Now, I think of people that are using
- 5:55Speaker 1: centralized AI open AI and know they don't really understand
- 6:00Speaker 1: the benefits of the decentralized AI or even AI agents.
- 6:06Speaker 1: So what would you say to those regular users that
- 6:10Speaker 1: they realize now they're relying on centralized AI every day
- 6:13Speaker 1: to work in their business or to do their job.
- 6:16Speaker 1: What's the advantage of the decentralized AI part and the
- 6:20Speaker 1: AI agents over just using a regular l M.
- 6:25Speaker 2: Yeah sure, yeah, Like all I described to early is
- 6:28Speaker 2: meant for developers. Are our primary customers kind of the
- 6:32Speaker 2: gate have been developers. That's why we I was a
- 6:36Speaker 2: little technical, but let's see, uh winding it back here.
- 6:40Speaker 2: The agents. The term agent is kind of we think
- 6:45Speaker 2: is the manifestation of all the promises of AI that
- 6:48Speaker 2: you've heard about if you if you look at open ai,
- 6:53Speaker 2: it's you know, it's it's a great tool. I use
- 6:55Speaker 2: it myself on It's the chat gipt. However, I think
- 6:59Speaker 2: it's kind of the convenience tool. It's too to help you,
- 7:03Speaker 2: help you cheat on your homework and speed up your
- 7:05Speaker 2: own personal processes. Maybe, however, we think that that is
- 7:11Speaker 2: just the first step of AI. Agents are kind of
- 7:16Speaker 2: the manifestation of automating different business processes. It's it's how
- 7:22Speaker 2: you create a business, It's how you interact with your followers,
- 7:25Speaker 2: interact with your customers. I compare touch you bet, open
- 7:29Speaker 2: ai too, kind of Yahoo dot Com in nineteen ninety nine.
- 7:33Speaker 2: It's a large learning model that kind of does everything everywhere,
- 7:37Speaker 2: all at once for everyone. However, you know, the rest
- 7:41Speaker 2: of the Internet needs to develop, you know, and we
- 7:45Speaker 2: see agents as the next step of like websites and
- 7:49Speaker 2: then apps, and then we think the next step evolution
- 7:52Speaker 2: of that is agents. Agents are going to be what
- 7:56Speaker 2: go get your news for you. They're gonna be what
- 7:59Speaker 2: trade on your back half, trade your stocks and bonds
- 8:02Speaker 2: the crypto, and they're going to be how you interact
- 8:07Speaker 2: with your followers in an automated way. On social media.
- 8:11Speaker 2: So I think agents are kind of the next major revolution.
- 8:16Speaker 2: And you know, large learning models were invested in heavily
- 8:21Speaker 2: for the last few years by Silicon Valley. However, kind
- 8:24Speaker 2: of deep seek I would say expose that and as
- 8:30Speaker 2: you know, not necessarily the be all end all. Someone
- 8:34Speaker 2: can in China can train their model that performs better
- 8:38Speaker 2: than years for sixty million dollars. That exposed the zero
- 8:41Speaker 2: sum game approach to large learning models. We think agents
- 8:45Speaker 2: are kind of something that you build a business around.
- 8:47Speaker 2: So for the everyday layman, I think I would say
- 8:52Speaker 2: look into agents because you can actually build small, medium
- 8:55Speaker 2: enterprises that displace a lot of the monopolies you see
- 8:58Speaker 2: around us today. Yeah, no, I love that.
- 9:01Speaker 1: Thank you for that comparison. And with all these enterprises
- 9:05Speaker 1: that are looking to get the proprietary data into AI,
- 9:09Speaker 1: how easy is that? You know, obviously with the help
- 9:13Speaker 1: of Gaya getting that set up. You know, I think
- 9:16Speaker 1: a lot of them are trying to figure out the
- 9:18Speaker 1: best way and how much handholding do you guys do,
- 9:22Speaker 1: and how quickly can you get all of that information
- 9:25Speaker 1: into the AI agent so that they've learned it fully
- 9:28Speaker 1: and you can start utilizing that data.
- 9:32Speaker 2: Yeah, the fact is that it's super simple. That's that's
- 9:35Speaker 2: our whole point. It's our framework is a very lightweight model.
- 9:39Speaker 2: You it depends how much data you have and how
- 9:42Speaker 2: much of a team you have to support it. But
- 9:45Speaker 2: I would say you can you can input an incredible
- 9:49Speaker 2: amounts of data, Like I'm talking about books and books
- 9:52Speaker 2: and reports and all kinds of pages and pages of
- 9:57Speaker 2: qualitative data within minutes. Right. It just depends. If you're
- 10:01Speaker 2: trying to upload an entire enterprises data set, then that's
- 10:05Speaker 2: probably going to take a few days, I would say,
- 10:08Speaker 2: but not that bad. It just depends on your you know,
- 10:12Speaker 2: the hardware and the human capital that you have access
- 10:16Speaker 2: to for if you're an enterprise. But really, look, we've
- 10:21Speaker 2: done pilots pilot agents for like Qualcomm, NASA, FIST ratings,
- 10:28Speaker 2: you know, La Fashion Week. We've got like over one
- 10:34Speaker 2: hundred and eighty what three pilots for the likes of Marsay,
- 10:37Speaker 2: JR Ritzu and layer Base and others. So yeah, there's
- 10:42Speaker 2: a lot of different case studies in there of how
- 10:45Speaker 2: how long it took, how what kind of agent was developed,
- 10:48Speaker 2: and kind of how we executed.
- 10:52Speaker 1: Yeah, no, I love that comparison. That you made of
- 10:56Speaker 1: like open aye as like yeahoo, I feel like they
- 11:01Speaker 1: It's crazy how much functionality you can have in that
- 11:06Speaker 1: first version of these lms, you know, compared to maybe
- 11:10Speaker 1: my generation wasn't around to see like that jumped from
- 11:13Speaker 1: like no Internet to the first search engine. But this,
- 11:17Speaker 1: it just seems incredible. And to think that that that
- 11:20Speaker 1: is actually the Yahoo and we're going to be going
- 11:23Speaker 1: so much farther, the functionality increase is going to be insane.
- 11:28Speaker 1: And would you make the comparison to say that like
- 11:31Speaker 1: every business, as they say now obviously for for decades now,
- 11:35Speaker 1: that every business needs to have a website, like every
- 11:37Speaker 1: business needs to have an AI agent as well.
- 11:41Speaker 2: Yeah, one hundred percent. I think that's the only thing
- 11:43Speaker 2: way forward. I think every business is going to be
- 11:49Speaker 2: wanting to explore agents. It's something that, like I said,
- 11:52Speaker 2: a new way to interact with your customers, your followers,
- 11:56Speaker 2: and it maybe improve if your digital business maybe completely bad,
- 12:00Speaker 2: tripport improve improve what you're serving. So like, for example,
- 12:04Speaker 2: the I encourage everyone to go build an agent and
- 12:08Speaker 2: look into what you can build. And I have a
- 12:10Speaker 2: whole list of ideas I've been suggesting crowds across the
- 12:13Speaker 2: world on UH you know what to what to do,
- 12:17Speaker 2: but essentially you know, for example, of an agent I
- 12:22Speaker 2: love the idea I love is I love an agent
- 12:24Speaker 2: where you know, essentially you tell it to go read
- 12:29Speaker 2: all the finance textbooks in the world. You tell it
- 12:34Speaker 2: to go follow Warren Buffett on Twitter, all the you
- 12:37Speaker 2: follow crypto Twitter, and then that agent could create sub
- 12:42Speaker 2: agents that they specifically follow equities, bonds, crypto commodities. And
- 12:49Speaker 2: then you tell it to go trade. Figure out a
- 12:52Speaker 2: medium conservative trading strid portfolio and go trade. That can
- 12:58Speaker 2: be done right now.
- 12:59Speaker 1: And uh.
- 13:00Speaker 2: The great thing is that's uh can even even scale
- 13:03Speaker 2: it so that you know, everyone who knows you and
- 13:06Speaker 2: knows that you're maybe the best agent trainer, UH can
- 13:10Speaker 2: follow you and pay you to trade their own portfolios.
- 13:15Speaker 2: That is a great multi million dollar.
- 13:18Speaker 3: Small business that'll you know, that displaces some of the
- 13:22Speaker 3: functions of the larger asset managers like a Charles Schwab
- 13:27Speaker 3: that you know kind of don't really they have very random,
- 13:31Speaker 3: basic robo traders.
- 13:34Speaker 2: That is, that's a great personalized business. That's what I'm
- 13:38Speaker 2: trying to get at where I think every business, large
- 13:41Speaker 2: business certainly should be looking at agents to keep up.
- 13:44Speaker 2: But I think what's more exciting is the ability for
- 13:49Speaker 2: the smaller medium sized businesses to dislodge them and perform
- 13:53Speaker 2: just as well. Uh, And that's really possible. Something like
- 13:58Speaker 2: that example is possible right now. What's missing is the
- 14:01Speaker 2: rails the infrastructure around agents. We think there's gonna be
- 14:05Speaker 2: more agents than humans in the next two to three years.
- 14:08Speaker 2: We've been building the infrastructure to service that. It's the
- 14:11Speaker 2: greatest population of unbanked on the planet, so it'll be
- 14:15Speaker 2: you know, we need KYC, we need AML banking wallets,
- 14:20Speaker 2: legal to serve all these agents that are going to
- 14:23Speaker 2: already coming about. So it's really exciting. And then you know,
- 14:28Speaker 2: to see how the way those agents interact with each
- 14:30Speaker 2: other on our behalf is also pretty exciting. Yeah, no,
- 14:35Speaker 2: it's fascinating.
- 14:36Speaker 1: And you give the example there of sort of taking
- 14:39Speaker 1: over the account manager for Charles Schwab and you mentioned,
- 14:42Speaker 1: you know, the legal I feel like AI agents can
- 14:45Speaker 1: at least do ninety nine percent of the grinding in
- 14:50Speaker 1: the legal industry as well. It feels like so many
- 14:54Speaker 1: different jobs are at risk, so to say, of AI
- 14:58Speaker 1: agents at least taking over the job or making it
- 15:01Speaker 1: almost obsolete. Do you think that people need to become
- 15:05Speaker 1: developers to not make their job become obsolete or just
- 15:10Speaker 1: learn how to manage AI agents? Where's where's the balance
- 15:13Speaker 1: in making sure that you still have a job when
- 15:15Speaker 1: AI agents are, you know, a bigger population than humans.
- 15:20Speaker 2: Yeah, So once again, I think there's two things in there.
- 15:23Speaker 2: I would say one is people going to lose jobs,
- 15:26Speaker 2: and two is you know, do they all need to
- 15:28Speaker 2: be developers the first the second question is that even
- 15:31Speaker 2: quicker to answer, it's no, they don't need to be developers.
- 15:34Speaker 2: That's the exciting part about this, you go. I was
- 15:36Speaker 2: just speaking to the Google folk. They I don't know
- 15:41Speaker 2: if you saw just last week they launched their kind
- 15:43Speaker 2: of no code coding AI. You can honestly make a
- 15:49Speaker 2: whole website or an agent or anything without being a
- 15:53Speaker 2: full developer. Obviously, if if you are a developer, it
- 15:55Speaker 2: definitely definitely helps and speeds things up. But I think
- 15:59Speaker 2: the rise of AI, large large learning models and agents
- 16:04Speaker 2: combined is making that skill less and less opaque for
- 16:11Speaker 2: the everyday man or woman. The other side of that is,
- 16:17Speaker 2: I think it's an interesting conversation that comes up every
- 16:20Speaker 2: time you bring up you know, you know, AI displacing
- 16:24Speaker 2: humans and taking over and rendering everyone jobless. I actually
- 16:30Speaker 2: think the opposite. I think that that conversation started because
- 16:35Speaker 2: Opening I didn't act as an open organization like I
- 16:38Speaker 2: was meant to be trained for about ten years or more,
- 16:41Speaker 2: and one day opened the flood gates and said, all right, great,
- 16:44Speaker 2: we know all of humanity's knowledge. We're good to go.
- 16:47Speaker 2: Pay us, pay us, We'll do everything for you. And
- 16:50Speaker 2: by the way, we can do everything, way overblown way,
- 16:54Speaker 2: it's way over formulated math kind of explanation of what
- 17:02Speaker 2: they can do and cannot do. For that matter, a
- 17:05Speaker 2: large line model from Opening EI. We were just in Korea.
- 17:08Speaker 2: It's our second largest market. It cannot service the Korean
- 17:11Speaker 2: market very well. Open as model is, like I said,
- 17:14Speaker 2: the Yahoo Dot comments does not interact well with the
- 17:19Speaker 2: Korean language. They have their own local ones that they're developing.
- 17:23Speaker 2: It does not interact well. It's not industry specific knowledge.
- 17:27Speaker 2: So if you're trying to figure out an actual chemistry
- 17:29Speaker 2: question from the industry, you're not going to get these
- 17:32Speaker 2: basic level answers. That's not good enough for you. So
- 17:35Speaker 2: I want to put to bed kind of the assumption
- 17:38Speaker 2: that Open Ai and Shachipi are just going to take
- 17:41Speaker 2: over the world because they themselves are not don't even
- 17:45Speaker 2: have enough proprietary knowledge to do so, and be their
- 17:48Speaker 2: capabilities are are not scalable and and are executable on
- 17:53Speaker 2: it for a small medium business. So let's get back
- 17:56Speaker 2: to like what you asked, like what can people do I,
- 18:00Speaker 2: Like I said with the finance agent example, this is
- 18:04Speaker 2: an excellent opportunity for small and media business to arise
- 18:08Speaker 2: and decentralize those larger, larger kind of monopolists and the behemits.
- 18:15Speaker 2: Like I said that the finance example is one it's
- 18:18Speaker 2: no ability for an everyday agent creator to go set
- 18:23Speaker 2: up something that dislodges potentially like a child Schwab or
- 18:27Speaker 2: JP Morgan. On the other hand, you can also create
- 18:31Speaker 2: a business that using agents in a unique way that
- 18:35Speaker 2: kind of dislodge, whether it's fashion, retail or social media
- 18:40Speaker 2: or you know, anything else that services yours regional and
- 18:46Speaker 2: local community and potentially international and it just depends on
- 18:50Speaker 2: your followers. We're kind of the YouTube for creating agents,
- 18:54Speaker 2: and as the YouTube, there's a whole host of there's
- 18:57Speaker 2: Mister Beast the mega level, but then there's also kind
- 19:02Speaker 2: of the small medium creators who are earning great income.
- 19:05Speaker 2: Let's created. It's dislodged the studio system of Hollywood, which
- 19:11Speaker 2: is like five or ten studios to millions of creators, right,
- 19:16Speaker 2: So we see this as the great dislodger of that.
- 19:20Speaker 2: It's decentralizing it and you know, spreading the wealth around
- 19:25Speaker 2: to more people. So I would encourage people to go
- 19:28Speaker 2: just figure out how to create agents. There's plenty of
- 19:31Speaker 2: auto create ones. We ourselves are in a joint venture
- 19:34Speaker 2: creating a not kind of a square space Shopify for
- 19:37Speaker 2: agents that allows you know, non creators to create them.
- 19:43Speaker 2: So there's gonna be a lot that pop up like that,
- 19:45Speaker 2: and they've already been a lot that popped up under
- 19:48Speaker 2: Google and others. So I think it's actually great for you.
- 19:51Speaker 2: Guman me No.
- 19:53Speaker 1: I love that, And I love that the decentralization allows
- 19:57Speaker 1: lowers the barriers to entry, allows anybody to do it,
- 20:00Speaker 1: whether they're in an emerging country or they're in a
- 20:03Speaker 1: first world country. And I'd love to learn more about
- 20:07Speaker 1: the decentralization or the blockchain part of it. And you
- 20:10Speaker 1: mentioned there's nodes and it can get into tokenization models
- 20:14Speaker 1: and stuff. How does the blockchain, you know, secure the
- 20:18Speaker 1: information or keep it private or what are the other
- 20:20Speaker 1: benefits in having it on a decentralized network.
- 20:24Speaker 2: Yeah, sure. Like like I mentioned earlier, we are an
- 20:27Speaker 2: inference network first and foremost. That allows your data is
- 20:33Speaker 2: stored on how are you like it. You can keep
- 20:35Speaker 2: it on your local, on prem on your devices, on
- 20:39Speaker 2: your hardware, you can keep it on AWS wherever you like.
- 20:43Speaker 2: You get to choose how it secured your data and
- 20:46Speaker 2: how you share it. The inference network purely allows you
- 20:52Speaker 2: to share the inferencing of your data unchained. I that's
- 20:57Speaker 2: at attribution, validation, and monetization as we call it. So
- 21:03Speaker 2: that's great.
- 21:03Speaker 1: And then with the there's the ability to have token
- 21:08Speaker 1: gated models or revenue models with actual cryptocurrencies as well.
- 21:12Speaker 2: Yeah, like I said, we are multi chain. We're kind
- 21:15Speaker 2: of a consensus layer that sits on top. We are
- 21:19Speaker 2: Ethereum first, but we will support other chains essentially. Uh,
- 21:25Speaker 2: the get dao governance is done on our on our
- 21:29Speaker 2: token as well as kind of the coordination of inference,
- 21:34Speaker 2: so you can stake any tokens, any type of crypto
- 21:38Speaker 2: on onto our a v s's using guigan. But we
- 21:43Speaker 2: are you know going to allow for settling of inference
- 21:48Speaker 2: payments and then selling of uh doing the on chain
- 21:52Speaker 2: transactions of inferencing on chain.
- 21:56Speaker 1: Mm hmm. I like that and I saw that the
- 22:00Speaker 1: day have recently reached four hundred thousand nodes on the network,
- 22:03Speaker 1: which is an incredible number. Do you have to if
- 22:07Speaker 1: you're setting up the AI agent, do you have to
- 22:10Speaker 1: run a node? Or can anybody that's using the agent's
- 22:13Speaker 1: run a node? Or what are the benefits of having
- 22:15Speaker 1: a node?
- 22:16Speaker 2: Yeah, we have. It's an interesting terminology for us. We
- 22:21Speaker 2: there's a few different types of nodes in our system.
- 22:24Speaker 2: There's a pure like a data node, like an agent
- 22:28Speaker 2: node that is kind of like the professors at Berkeley
- 22:32Speaker 2: who are you know, have uploaded their ip so anyone
- 22:36Speaker 2: can upload their knowledge and there the agent the way
- 22:40Speaker 2: it interacts onto their own agent node. Every agent is
- 22:46Speaker 2: a node on our platform, so that is you know
- 22:51Speaker 2: that you can be shared and monetized on our system.
- 22:54Speaker 2: Like I mentioned, there's kind of validator nodes, people who
- 22:58Speaker 2: steak and validate and maintain the network. Uh, there's compute nodes.
- 23:05Speaker 2: There's a bunch of deep in providers like foul coin.
- 23:09Speaker 2: I am at deepen x others who are who have
- 23:13Speaker 2: massive cap x and are running a lot of compute.
- 23:15Speaker 2: We're we're helping them match the customers for their compute
- 23:19Speaker 2: from developers these agents, uh, these kind of knowledge agents.
- 23:25Speaker 2: So that's another kind of user base. And then the
- 23:28Speaker 2: last kind of uh type of node is the developer nodes.
- 23:33Speaker 2: They these guys are people who build apps adaps on
- 23:36Speaker 2: the network and they point to these the knowledge and
- 23:40Speaker 2: compute nodes uh in order to power their apps. So yeah,
- 23:46Speaker 2: there's a few user profiles, but the terminode is a
- 23:49Speaker 2: little kind of uh overuse it to maybe, but it's
- 23:54Speaker 2: a it's has a few different meanings on our network.
- 23:56Speaker 1: Definitely. No, Well, it's great to see the traction and
- 24:00Speaker 1: I feel like over the next year it's probably gonna
- 24:02Speaker 1: explode into the millions of nodes and hopefully, you know,
- 24:05Speaker 1: the more decentralization the better as well.
- 24:09Speaker 2: Yeah, I think it's been a wild ride because we're
- 24:13Speaker 2: we're just about the year mark of officially launching, so
- 24:17Speaker 2: it's you know, we had about three hundred thousand nodes
- 24:21Speaker 2: at early April and now where we had a four
- 24:24Speaker 2: hundred thousand even quicker, So it's getting quicker.
- 24:28Speaker 1: Exponantial growth, just like the Knowledge game, but from AI.
- 24:32Speaker 1: It's in the trajectory of the functionality of these AI agents.
- 24:36Speaker 1: It's going to pick up. I don't know if humans
- 24:40Speaker 1: will be able to keep up with and it's already
- 24:42Speaker 1: hard to keep up with the crypto news, never mind AI.
- 24:45Speaker 2: So there'll be more agents than humans very quickly. Yeah, definitely.
- 24:49Speaker 1: And you mentioned in the beginning with the guys at
- 24:52Speaker 1: Berkeley and the researchers, they have their own domain. I
- 24:55Speaker 1: saw that there's like domain names that you can get
- 24:57Speaker 1: as well.
- 24:57Speaker 2: How does that work?
- 24:58Speaker 1: And you know, what's the benefit of having a domain
- 25:02Speaker 1: so that you can people you can direct people to
- 25:04Speaker 1: the agent on its own domain.
- 25:06Speaker 2: Yeah, I mean, it's good to own your own domain.
- 25:09Speaker 2: It's just like a website. We're actually partnering with n
- 25:11Speaker 2: S on that, but we yeah, our domains, like I said,
- 25:16Speaker 2: represent it can represent just one agent, just the one
- 25:21Speaker 2: that you built, or you know, we can get you
- 25:24Speaker 2: an Ashton agent, right and that can be accessed by
- 25:29Speaker 2: you know, actually a guy in AI. There's also enterprises
- 25:32Speaker 2: like Berkeley mentioned institutions that have multiple nodes underneath it.
- 25:37Speaker 2: So for them it acts as like the gateway to
- 25:41Speaker 2: all the agents and nodes that they're hosting. So you
- 25:45Speaker 2: know what, it's just like we're at the beginning of
- 25:48Speaker 2: the Internet and you want to own your domain before
- 25:51Speaker 2: the Internet, guess right, So definitely and some some of
- 25:55Speaker 2: them it's the RGD service. The domain service has been
- 26:00Speaker 2: exploded in popularity. We've done a couple of hundred grand
- 26:04Speaker 2: of revenue there but already. But we I think it's
- 26:09Speaker 2: it's a mix of between speculators and actual true business
- 26:14Speaker 2: owners and others and our enterprises. So it's exciting.
- 26:19Speaker 1: Yeah, that is exciting and what you know lies ahead
- 26:23Speaker 1: for for you and the team at Gaya in building
- 26:25Speaker 1: this out and as AI functionality just exponentially grows. As
- 26:30Speaker 1: we're saying, you know, for the rest of twenty twenty five,
- 26:33Speaker 1: is there anything public that you're planning to release on
- 26:36Speaker 1: the roadmap or you know, how do you continue to grow?
- 26:39Speaker 2: Yeah, I think we have. We have a lot of
- 26:41Speaker 2: awesome JVS partner lined up. We you know, we we've
- 26:46Speaker 2: integrated with a lot of the wallet providers, so that's
- 26:49Speaker 2: gonna lead to true proliferation of you know, being able
- 26:52Speaker 2: to use your agents in any way like for example,
- 26:56Speaker 2: Consensus is on our cap table Meta mask. We're trying
- 27:00Speaker 2: to launch a metal mass market agent marketplace with them,
- 27:04Speaker 2: the transportable across all kinds of wallets. So we've already
- 27:09Speaker 2: integrated with finance wallet, okay x wallet, base coin Base
- 27:14Speaker 2: where we're very we have a lot of partnership with
- 27:17Speaker 2: Base or maybe s is on there. We're doing a
- 27:21Speaker 2: partnership with moon Pay on multiple things, but mostly account obstruction.
- 27:26Speaker 2: You want to create off on ramp for agents. So
- 27:30Speaker 2: there's a lot of cool kind of infrastructure that's gonna
- 27:33Speaker 2: be deployed that already fits within user their everyday tools
- 27:39Speaker 2: that they use. But then there's you know the some
- 27:43Speaker 2: of the jvs like that square space for agents that
- 27:47Speaker 2: I mentioned that we're doing with a very large enterprise
- 27:52Speaker 2: that is going to allow to really blow up in
- 27:55Speaker 2: the door of floodgates for any any non debt to
- 27:57Speaker 2: create it. We have a potential hacketon with Google and
- 28:02Speaker 2: a few others coming up. But yeah, look for we're
- 28:05Speaker 2: going to launch our network at some point this year.
- 28:08Speaker 2: It's going to lead to us being able to actually
- 28:11Speaker 2: execute on some of the transactions that I mentioned, and
- 28:17Speaker 2: we got a big year ahead of us regardless of
- 28:19Speaker 2: where crypto parkets sit or don't sit. We're not based
- 28:23Speaker 2: on the volatility of the markets because we've been building
- 28:27Speaker 2: and we truly believe AI and agents are the next
- 28:32Speaker 2: big adoption mechanism of Web three. In general, I think
- 28:37Speaker 2: to date the majority of adoptions still remains, you know,
- 28:42Speaker 2: financial services and wealth past management. But we think that,
- 28:48Speaker 2: like I mentioned earlier in my kind of narrative, we
- 28:52Speaker 2: strongly believe that we're up against a vertical cliff in
- 28:57Speaker 2: AI where the only way to surmount it is to
- 29:01Speaker 2: Web three. So we think that even silicon value will
- 29:04Speaker 2: start to kind of come back to Web three rails,
- 29:08Speaker 2: and so we see ourselves as a kind of the
- 29:10Speaker 2: cloud infrastructure provider in the nineties for all the websites
- 29:15Speaker 2: that came after it, websites and apps that came after it.
- 29:18Speaker 2: So it's exciting times for both Web three and US. Yeah.
- 29:22Speaker 1: No, that's very exciting, and I'm looking forward to to
- 29:25Speaker 1: that launch and I'm going to check out the AI agents,
- 29:30Speaker 1: see what I can create myself, and also the domain names,
- 29:33Speaker 1: try to secure some good ones in my own. What's
- 29:36Speaker 1: the best way to get started on both of those things?
- 29:40Speaker 2: Yeah, go find us if you're a developer file you
- 29:44Speaker 2: can go open source, go clone down our stack, play around,
- 29:49Speaker 2: break it, fix it, you know, give us feedback. That's
- 29:53Speaker 2: the great thing here. Unlike a lot of projects in
- 29:55Speaker 2: the space, we're pretty open book. You can or run
- 29:59Speaker 2: at Linux model, so you can go to play around
- 30:01Speaker 2: and if you don't like it, give us feedback. And
- 30:03Speaker 2: there's over four hundred thousand known there, so you can
- 30:06Speaker 2: rest a shirt. We've received quite a bit of feedback already,
- 30:09Speaker 2: luckily mostly positive and yeah, I mean if you're not
- 30:13Speaker 2: a non down like so check with our updates, follow
- 30:16Speaker 2: our trologram, join our discord, follow our socials for updates.
- 30:21Speaker 2: Will definitely be interacting with the community in a lot
- 30:24Speaker 2: of different ways in the coming months. Sounds great to Shank.
- 30:28Speaker 1: I really appreciate the insights into AI agents and the
- 30:33Speaker 1: fact that it needs to be on Web three for
- 30:35Speaker 1: this next stage of adoption. That's exciting to all the
- 30:38Speaker 1: crypto junkies as well. So I love what the team's
- 30:43Speaker 1: working on. Wishing you and everyone a guy all the
- 30:46Speaker 1: best moving forward. I'm going to leave those social links
- 30:48Speaker 1: and the dev links in the show notes below as well,
- 30:51Speaker 1: and we would love to follow up again in the
- 30:53Speaker 1: near future.
- 30:54Speaker 2: Awesome, Thanks so much, Ashan,