Latest / Crypto Coin Show / Inside Fetch.ai & ASI: Building the Future of Autonomous Agents | Blockchain Interviews
Transcript
- 0:00Speaker 1: I'm Ashton Addison from the Crypto Coin Show, and today
- 0:03Speaker 1: on Blockchain Interviews, we have Humayan Shake, CEO and co
- 0:06Speaker 1: founder of Fetch AI here to talk about decentralized AI,
- 0:11Speaker 1: autonomous agents, and the open intelligent machine economy that is
- 0:16Speaker 1: coming human. Thank you so much for taking the.
- 0:19Speaker 2: Time, appreciate it. Thank you. It's a pleasure to be here.
- 0:23Speaker 1: Likewise excited to dive into your expertise in AI.
- 0:28Speaker 3: Over the past few years.
- 0:29Speaker 1: It's exploding and it's hard for people to catch up,
- 0:32Speaker 1: even people that are in crypto following the news every day.
- 0:35Speaker 1: AI is moving so quickly, and I know you've been
- 0:38Speaker 1: studying it for a long time. I'd love to start
- 0:41Speaker 1: out with maybe a little bit on your background in
- 0:44Speaker 1: studying AI and how you got involved with fetch and
- 0:48Speaker 1: then we can talk about all things AI.
- 0:51Speaker 2: Yeah, that's great. So my background computer science engineering and
- 0:59Speaker 2: got into through gaming quite early on, probably you know.
- 1:04Speaker 2: That was my first kind of love into gaming, building games.
- 1:08Speaker 2: That's how I met Demessavis who was deep the deep
- 1:12Speaker 2: Mind founder, and we got connected two thousand and probably five,
- 1:18Speaker 2: two thousand and six. Then we kind of started discussing
- 1:22Speaker 2: two thousand and eight nine, we set up a deep Mind,
- 1:26Speaker 2: which exited to Google in twenty thirteen, and I was
- 1:31Speaker 2: looking after the commercialization of AI, which means, you know,
- 1:34Speaker 2: how are we going to take this technology, how we're
- 1:36Speaker 2: going to go to the market with it, and how
- 1:39Speaker 2: are we going to make it into revenue generating kind
- 1:43Speaker 2: of technology. Looking at it the technology itself, and then
- 1:46Speaker 2: finding ways to adapt to the market. So that's that's
- 1:52Speaker 2: then twenty thirteen. Then I, when we came out of Google,
- 1:56Speaker 2: it was quite clear commercialization of that will not be
- 2:00Speaker 2: that easy unless you're a big company, a big enterprise,
- 2:04Speaker 2: because you have all the data, you have all the
- 2:07Speaker 2: relevant tools. So my focus then became how can you
- 2:11Speaker 2: bring it to the masses and what a real component
- 2:15Speaker 2: would enable you to do that, And it became quite
- 2:18Speaker 2: clear it will be like agentic systems because agents, you
- 2:23Speaker 2: can bring agents, individuals can interact with agents, you can
- 2:27Speaker 2: break out the tasks to use agents to fulfill them,
- 2:31Speaker 2: and that's how the whole thing will kind of evolve.
- 2:34Speaker 2: And on the other side, it was the language which
- 2:36Speaker 2: was going to be the key, because the interface was
- 2:39Speaker 2: going to be language and the ability to execute will
- 2:43Speaker 2: be the agents. So that's how I got into the
- 2:46Speaker 2: agentic system. So twenty eighteen, twenty nineteen, we started up
- 2:49Speaker 2: Fetch with that premise. So we've been around a while,
- 2:54Speaker 2: as you probably know, and people used to think when
- 2:59Speaker 2: we talk about agents, it was crazy, not going to happen.
- 3:03Speaker 2: It's just a narrative which we created. But I think
- 3:07Speaker 2: they stand corrected on this one, the people who were
- 3:11Speaker 2: the naysayers. We've always tried to stay ahead of the
- 3:16Speaker 2: tech and you know, kind of create the pathway. So
- 3:21Speaker 2: if you think about agents, we were the first project.
- 3:24Speaker 2: Deep Mind was the first company which kicked off this
- 3:27Speaker 2: whole AI phase. We then did a first merger in crypto,
- 3:31Speaker 2: the biggest one we could do, you know, as pluses
- 3:35Speaker 2: and minuses, but time will tell, and then you know,
- 3:39Speaker 2: we kind of we're trying to build, you know, ahead
- 3:42Speaker 2: of people and something which brings utility to decent life space.
- 3:48Speaker 2: Really that's kind of an overall journey.
- 3:51Speaker 1: And here we and here we are, and yeah, i'd
- 3:54Speaker 1: love to dive into that merger in a little bit
- 3:56Speaker 1: about as I and you know, what's the reasoning behind
- 4:00Speaker 1: that and why it's so valuable. I want to ask
- 4:02Speaker 1: one more question about agentic AI and you know, Since
- 4:06Speaker 1: twenty eighteen, with the uprising of LMS, a lot of
- 4:11Speaker 1: you know, regular everyday people have started to use language
- 4:15Speaker 1: to communicate with with agents per se. But I feel
- 4:18Speaker 1: like they people still don't understand the full functionality and
- 4:23Speaker 1: capacity of what it is to become because I feel
- 4:25Speaker 1: like there's a lot of functions that we still and
- 4:28Speaker 1: we still have we have to ask AI and we
- 4:30Speaker 1: have to do it ourselves for the most part, and
- 4:32Speaker 1: I feel like in the coming years all of that
- 4:36Speaker 1: will be taken care of and it's going to be
- 4:38Speaker 1: so much bigger than it still is now.
- 4:41Speaker 2: Yes, I think we're just at the stop. What it's
- 4:45Speaker 2: not even the stop, it's you know, just before the stop.
- 4:50Speaker 2: But we're on the journey. So so what what I
- 4:53Speaker 2: mean you, I'm sure you have talked about agents so
- 4:56Speaker 2: many times? Do you actually have an agent? Did you? Eight? One?
- 5:00Speaker 2: And how? Yeah?
- 5:02Speaker 1: Yeah, I think there's a lot of people that, you know,
- 5:04Speaker 1: there's a lot of discussion around AI agents. I think
- 5:07Speaker 1: for the most part, people are relying on these large
- 5:10Speaker 1: language models, and I've personally used a few AI agents
- 5:14Speaker 1: here and there in telegram things like that, but even
- 5:18Speaker 1: being involved in crypto, I don't think I'm taking advantage
- 5:22Speaker 1: of it as much as I could be, and maybe
- 5:24Speaker 1: the barriers to entry and understanding what are the things
- 5:27Speaker 1: possible and how do I actually get started doing that
- 5:30Speaker 1: to help me run my business or my everyday life.
- 5:33Speaker 1: I think a lot of people it needs to be
- 5:35Speaker 1: broken down a little bit more.
- 5:37Speaker 2: Still. Yeah, so that's the kind of tools you're bringing,
- 5:42Speaker 2: which is how can a person who is not a coder,
- 5:45Speaker 2: who doesn't want to be too involved in technology, can
- 5:50Speaker 2: can engage with technology. I think chat GPT did a
- 5:53Speaker 2: great job where it gave this LLM models in the
- 5:58Speaker 2: hand of everybody and they started using it, and they
- 6:00Speaker 2: started using it because they didn't need any technical help, right.
- 6:04Speaker 2: So that's what's going to happen with agents as well.
- 6:07Speaker 2: You will have your own agent, you will be represented
- 6:10Speaker 2: by your agent. You'll have two lives effectively. One is
- 6:13Speaker 2: the professional life, one is the personal life. And then
- 6:17Speaker 2: the agent will take a input from you or your intent,
- 6:23Speaker 2: your command, your demand, you know, all of those things,
- 6:28Speaker 2: and it will interpret that into actions and then carry
- 6:32Speaker 2: out those actions. So the first step in commercialization where
- 6:35Speaker 2: people can actually start using it is provide something very
- 6:39Speaker 2: simple just like Chadgibt. You should be able to set
- 6:42Speaker 2: up your agent, and then you should be able to
- 6:45Speaker 2: communicate with your agent. No different to what you do
- 6:48Speaker 2: with Chadgibt, because that's just language, and then you can
- 6:51Speaker 2: ask it to do things and then it goes and
- 6:55Speaker 2: interacts with other agents to actually carry out you know,
- 6:59Speaker 2: it could be just a textual communication, it could be
- 7:01Speaker 2: more execution, It could be generation or something. It could
- 7:06Speaker 2: be any of those things, and that's when you start
- 7:10Speaker 2: seeing the traction. So my belief is anybody who has
- 7:15Speaker 2: a website or a social media page or a professional
- 7:20Speaker 2: page like LinkedIn page, and all of that will become
- 7:26Speaker 2: attached to an agent. So all of that will become agentic.
- 7:30Speaker 2: And if you think about just that, it's a huge,
- 7:34Speaker 2: huge market, so you actually see what's coming. So all
- 7:39Speaker 2: of those will be agents, and you know, you could
- 7:42Speaker 2: have every company will have an agent or multiple agents.
- 7:44Speaker 2: So some will be internal, some would be eternal. They
- 7:47Speaker 2: will be represented by agents, and individuals are the same.
- 7:51Speaker 2: But let's say if I play too parallel scenarios, if
- 7:57Speaker 2: you think about what happened with web every went off
- 8:01Speaker 2: and started creating a website was very difficult in the
- 8:03Speaker 2: beginning and became very easy now, so you can deploy it,
- 8:07Speaker 2: click it, click a few clicks, and you deploy it.
- 8:09Speaker 2: That's going to happen to the agents. Each website will
- 8:14Speaker 2: become agentic, so it's where you have a website, it
- 8:17Speaker 2: will be an agent. So when the reason why that
- 8:20Speaker 2: would be the case is you would have you know,
- 8:24Speaker 2: the websites give you information, but it's one way you interact.
- 8:28Speaker 2: You do the actions. Agents will be more interactive, more intelligent,
- 8:33Speaker 2: more communicative. You can have different things they can do,
- 8:37Speaker 2: so it will become more dynamic, it will become alive.
- 8:40Speaker 2: So that's that's kind of the parallel you can imagine.
- 8:43Speaker 2: Then comes the e commerce, which was on the website.
- 8:47Speaker 2: We built e commerce solutions. The same will happen with
- 8:49Speaker 2: the agents because they will need to interact, they will
- 8:52Speaker 2: need to transact. So that's the journey we've just started.
- 8:56Speaker 2: And all of that is needs to kind of not
- 8:59Speaker 2: be fitted back into Web two. It needs to be
- 9:02Speaker 2: AI first. Now it's not web first, it's AI first.
- 9:06Speaker 2: So what does that mean. What it means is that
- 9:09Speaker 2: you should be doing the work. You need the large
- 9:12Speaker 2: language models or AI to actually do the sum of
- 9:15Speaker 2: the work and then bring these solutions to you and
- 9:18Speaker 2: that will change the game quite a lot over the
- 9:22Speaker 2: next few years. And that's how we're going to start
- 9:25Speaker 2: getting using AI a lot more than what we do today.
- 9:28Speaker 2: So we're not there, we're just starting the.
- 9:30Speaker 1: Journey, definitely, And what needs to be done in do
- 9:35Speaker 1: you teach the agents you know, how to take those
- 9:40Speaker 1: actions based on the command?
- 9:43Speaker 3: And like, how exactly do we get to that point?
- 9:46Speaker 2: Yeah, I don't think you're going to need to teach
- 9:49Speaker 2: them as such. You carry out the actions, they will
- 9:53Speaker 2: carry out the actions for you, and that will result
- 9:56Speaker 2: in the learning will happen in the background, So you
- 10:01Speaker 2: don't really I don't feel we need we will need
- 10:04Speaker 2: to do anything. I think you'll start from small tasks.
- 10:07Speaker 2: You'll guide the agents to do those tasks, and then
- 10:11Speaker 2: suddenly they will start learning it. They will start doing it.
- 10:14Speaker 2: Of course, in parallel you're building the technologies, you're building
- 10:18Speaker 2: new models which are much better. They know what they're doing. So,
- 10:21Speaker 2: for example, Fetch has a model which is very tuned
- 10:26Speaker 2: for agentic systems. What that means is when you go
- 10:29Speaker 2: to Chad Gibt, you say, hey, you know, do this
- 10:31Speaker 2: for me? It goes and does that textual form, graphic
- 10:37Speaker 2: digital form. And then but when you say to it, okay,
- 10:40Speaker 2: you know, now connect me to this person's agent. You
- 10:43Speaker 2: can't connect to agents, right, It's all one model encompassing
- 10:47Speaker 2: which goes out to do different tools. But these tools
- 10:50Speaker 2: have done for now, and if you think of them
- 10:53Speaker 2: to be more alive and dynamic and intelligent, they will
- 10:56Speaker 2: be the agents. So what we have for example today
- 11:00Speaker 2: is and you can you can actually your listeners can
- 11:03Speaker 2: go and test it out on a SI one. We
- 11:05Speaker 2: have what we call is an agentic LLLM. What you
- 11:09Speaker 2: can say is, hey, I need to do something, and
- 11:12Speaker 2: it will go and find the agents that can actually
- 11:14Speaker 2: do it. So people outside of the LLLM can build
- 11:18Speaker 2: the agents and deploy them, and you can actually find
- 11:23Speaker 2: them by going onto a normal textual conversational interface.
- 11:30Speaker 3: That's really interesting. I'd love to know more about the ASI.
- 11:35Speaker 1: You know, I believe it's artificial superintelligence merger and what
- 11:39Speaker 1: was the you know, the reasoning and the value behind
- 11:42Speaker 1: making this one big unit.
- 11:47Speaker 2: Yeah. So, so if you look at what's happening in
- 11:50Speaker 2: the world, you don't just you don't just build models
- 11:54Speaker 2: because you need compute to build those models. You need
- 11:57Speaker 2: data to build those models. Then you'll also need to
- 12:01Speaker 2: commercialize these models that otherwise, what are we building them for?
- 12:05Speaker 2: They need to do something. That's the application layer. So
- 12:08Speaker 2: if you if you work out this AI stack, the
- 12:13Speaker 2: stack starts from right on the top, there is an
- 12:15Speaker 2: application layer. How do you build these applications? If that's agents,
- 12:21Speaker 2: you can't underneath that. These agents need tools and needs integrations,
- 12:27Speaker 2: and these agents need learning methodologies algorithms. So that's that's
- 12:33Speaker 2: your machine learning models that need to be built. Then
- 12:37Speaker 2: you come down further to do all of that, you
- 12:40Speaker 2: need really data because without data you can't so you
- 12:45Speaker 2: need data tools. And then if you go further down,
- 12:50Speaker 2: these models need to be deployed somewhere. They need to build.
- 12:53Speaker 2: They need to have compute, so you need a compute network.
- 12:56Speaker 2: As you saw, I mean n Video is shot up
- 12:59Speaker 2: became one of the most valuable companies in the world
- 13:02Speaker 2: because everybody understands this AI phase is going to need
- 13:07Speaker 2: a lot of compute, So you need the compute. So
- 13:11Speaker 2: owning the whole tech stack or actually having an integration
- 13:15Speaker 2: between the tech stack makes it a lot more efficient.
- 13:18Speaker 2: And if you look at especially the crypto part of
- 13:21Speaker 2: the world. There is no project which delivers that whole
- 13:27Speaker 2: tax stack. There is no project where you can go
- 13:31Speaker 2: right from the I'm going to build the agent. I'm
- 13:33Speaker 2: going to ask the agent to deploy a machine learning
- 13:35Speaker 2: model on this compute, and I'll run it, and I
- 13:38Speaker 2: will make my agent available and people will be able
- 13:41Speaker 2: to interact with it. Does not exist. It's the whole
- 13:45Speaker 2: tax stack. And then on top of that, you need
- 13:47Speaker 2: a lot of research to keep building these solutions for
- 13:50Speaker 2: the next generation of AI. You need to stay on
- 13:53Speaker 2: top of the So taking that all those projects, which
- 13:59Speaker 2: includes fetcha Hind, Singularity, Ocean, and Kudos, putting them together
- 14:05Speaker 2: in a way that the tech stack makes sense and
- 14:08Speaker 2: is much more efficient and it's much more encompassing as
- 14:14Speaker 2: compared to anybody else and uses one token ultimately to
- 14:18Speaker 2: use all of those services, rather than going one token,
- 14:21Speaker 2: convert to another token, another change and that the other.
- 14:25Speaker 2: You know, that's if we want to build a solution
- 14:28Speaker 2: which world is going to use, we need to do
- 14:30Speaker 2: it efficiently, we need to do it quickly, and we
- 14:32Speaker 2: need to do it in a way that people use
- 14:35Speaker 2: it to bring the utility. So that was the reason
- 14:37Speaker 2: for the merger.
- 14:39Speaker 3: Yeah, no, it's really exciting, and it's.
- 14:43Speaker 1: It's you know, history in the small industry that crypto
- 14:47Speaker 1: is the small market capitalization compared to the world that
- 14:52Speaker 1: they always say, you know, these blockchain companies, they're on
- 14:55Speaker 1: the same we're on the same team. You know, it's
- 14:56Speaker 1: such a small industry. We need to work together to
- 14:59Speaker 1: grow in more of the mainstream. So I'm glad to
- 15:02Speaker 1: see that. You know, it's not competition, it's cooperation. So
- 15:06Speaker 1: it's really exciting.
- 15:07Speaker 2: That's that's really the premise of the whole thing, and
- 15:11Speaker 2: that's what we did that for. And actually, you know,
- 15:16Speaker 2: the more the more we are working together, the more
- 15:18Speaker 2: we see that the tech stack makes sense.
- 15:21Speaker 3: Yeah, that's great.
- 15:23Speaker 1: I would love to talk a little bit more about
- 15:25Speaker 1: decentralized AI and maybe the blockchain part of it, but
- 15:29Speaker 1: there's been this, uh, this battle through calling out you know,
- 15:36Speaker 1: open AI that it's not really open it's more centralized AI,
- 15:39Speaker 1: and decentralized AI has other advantages. Can you talk about
- 15:44Speaker 1: the importance if there is, of having actual decentralized AI
- 15:49Speaker 1: versus centralized.
- 15:54Speaker 2: It's it's a misnomer because I think people use it
- 15:58Speaker 2: just to launch a new project. Mostly it's decentralized AI.
- 16:03Speaker 2: What is decentralized AI? We need to actually define it.
- 16:08Speaker 2: We need to actually understand that what components are decentralized.
- 16:13Speaker 2: So so if I if I kind of break it
- 16:16Speaker 2: down into what what does it mean to have decentralized AI.
- 16:21Speaker 2: AI is not one big thing right in if you
- 16:24Speaker 2: if you think of AI, AI is made up of models.
- 16:29Speaker 2: AI is made up of smaller models, large models, large
- 16:32Speaker 2: language models, large other types of models, specialist models. They
- 16:37Speaker 2: then interact with each other. How do they interact with
- 16:40Speaker 2: each other? I mean they could be agenic systems, they
- 16:42Speaker 2: could be just integrations with APIs so that they can
- 16:45Speaker 2: actually interact with each other. So when I when I
- 16:49Speaker 2: think of decentralized AAI, what I think is that if
- 16:53Speaker 2: somebody went and pressed the button, can they turn it off?
- 16:57Speaker 2: So well, so if I take that theme, then now
- 17:04Speaker 2: we have so many llms, right, so if you don't
- 17:06Speaker 2: have chagbut you'll have something else. We have our own.
- 17:10Speaker 2: How were they trained? They were trained on a decentralized
- 17:13Speaker 2: compute network. So if one company says we don't want
- 17:17Speaker 2: you to train this model and we're going to turn
- 17:20Speaker 2: you off, it doesn't matter because this is decentralized and
- 17:25Speaker 2: no one person can turn it off. Now, when we
- 17:28Speaker 2: serving inference and somebody doesn't like the inference, can we
- 17:31Speaker 2: stop the inference being served? No, because it's being served
- 17:36Speaker 2: via a decentralized system. Did we train it on decentralized Yes,
- 17:41Speaker 2: we did. Do we serve it via decentralized Is it
- 17:44Speaker 2: posted on decentralized compute? That in itself makes it a
- 17:48Speaker 2: little bit more decentralized than anything else. Now, and you
- 17:54Speaker 2: can train it by using and interacting with multiple participants
- 17:59Speaker 2: and in it with multiple inclusive data from different participants.
- 18:04Speaker 2: That's also adding to that. Now, the biggest part portion
- 18:08Speaker 2: is when you go out in the market and do
- 18:10Speaker 2: something with it. Is that decentralized and what that means
- 18:13Speaker 2: is you can have your own agent, you can be
- 18:16Speaker 2: running your own AI models. We can do all of that.
- 18:19Speaker 2: So there is value, but the value comes at a price.
- 18:23Speaker 2: That decentralized value. The price is the cost. The efficiency
- 18:28Speaker 2: not as efficient costly, It's not easy to do. So
- 18:32Speaker 2: all of those things and all of those components need
- 18:35Speaker 2: you really need to have a conviction that decentralized AI
- 18:41Speaker 2: is what you want. But most people don't care. Most
- 18:44Speaker 2: people care about utility. So if you don't provide them
- 18:48Speaker 2: utility and you just keep saying, oh, it's decentralized, nobody cares, right,
- 18:54Speaker 2: Only crypto community cares. And that's a small committee, right,
- 18:57Speaker 2: So what we need to do and actually even within
- 19:00Speaker 2: the crypto community, only a very small actual uh you know,
- 19:05Speaker 2: actual people who do care, and the rest of them
- 19:07Speaker 2: are actually doing it as a lip service because that's
- 19:10Speaker 2: our unique selling point. We are decentralized. No, that's not
- 19:14Speaker 2: That should not be the central selling point. That should
- 19:18Speaker 2: not be the unique selling point. We need to make
- 19:21Speaker 2: utility out of it. First, people should use it. Then
- 19:25Speaker 2: it's DEPLI then it makes sense. So I have kind
- 19:30Speaker 2: of mixed opinions on it. I think we need to
- 19:33Speaker 2: have decentralized AI and when AGI comes, you want to
- 19:37Speaker 2: have all those components, and we're proponents of that. I'm
- 19:40Speaker 2: very strong proponent of it. But what I am not
- 19:45Speaker 2: going to support is a you know, is the argument
- 19:48Speaker 2: that everything is okay if it's decentralized, it's not. If
- 19:52Speaker 2: it's not, if it's not useful, then it's not okay.
- 19:55Speaker 3: Mm hmm.
- 19:56Speaker 1: Yeah, that's a great perspective, and with businesses that are
- 20:01Speaker 1: you know, there's so many of these huge corporations that
- 20:04Speaker 1: have a lot of proprietary information for them building their
- 20:07Speaker 1: own agents that you know, they can put their insider
- 20:11Speaker 1: corporate information in there to be more efficient without having
- 20:15Speaker 1: to share it with open AI for example, or other corporations.
- 20:19Speaker 1: Is that does that require decentralized AI or just private AI.
- 20:27Speaker 2: You're always going to have a mix of it. You're
- 20:30Speaker 2: going to have private AI, You're going to have mixed
- 20:32Speaker 2: always going to have mixed mixture. Because people don't want
- 20:35Speaker 2: to share data. People don't want to share data. I mean,
- 20:38Speaker 2: that's that's the point we have all seen now over
- 20:41Speaker 2: the last how many years. Data is not going to
- 20:44Speaker 2: be shared by people. And the only thing you can
- 20:48Speaker 2: do is you can train AI model in a decentralized
- 20:50Speaker 2: way where people can participate in training of the model.
- 20:54Speaker 2: But even that is very difficult for people to convince
- 20:57Speaker 2: to do because you know, if I'm spending money and
- 21:00Speaker 2: I want to make something out of it, I'm not
- 21:03Speaker 2: just doing it for sake of pulling it kind of
- 21:06Speaker 2: decentralized or not centralized or whatever. I'm doing it for
- 21:10Speaker 2: actually generating value. Economic value and that's gonna yeah. So
- 21:16Speaker 2: I think there's good and bad components of both sides,
- 21:19Speaker 2: which we'll see kind of playing out over the next
- 21:22Speaker 2: few years.
- 21:24Speaker 1: I actually read that some of the industries, you know,
- 21:27Speaker 1: not finance, not crypto related, are outpacing like Silicon Valley,
- 21:33Speaker 1: in AI adoption, like healthcare, these other industries.
- 21:37Speaker 3: I feel like AI is going to tap into all
- 21:39Speaker 3: of them. Have you seen this?
- 21:43Speaker 2: AI is going to be part of everything, So that's
- 21:46Speaker 2: a given. But but you know what is quite interesting
- 21:51Speaker 2: is like I don't know if you know about the
- 21:55Speaker 2: payment payment app called Paisa, which which became very popular
- 22:01Speaker 2: in Africa. The reason why that there was no existing infrastructure.
- 22:05Speaker 2: So places or sectors or vertical sectors which do not
- 22:12Speaker 2: have too much new technology already, they can easily go
- 22:18Speaker 2: AI first and they can start building solutions with AI first.
- 22:23Speaker 2: And that's where the agentic systems come really well, because
- 22:27Speaker 2: even if you have legacy systems, you can very easily
- 22:30Speaker 2: put an agent on it and you can join this
- 22:33Speaker 2: new AI race because it can suddenly bring you into
- 22:37Speaker 2: It's a retrofitting mechanism. Right As much as agents are,
- 22:41Speaker 2: AI based agents can be very useful even if you
- 22:44Speaker 2: don't retrofit them on legacy systems. But they actually they
- 22:50Speaker 2: act as a layer between the legacy system and the
- 22:52Speaker 2: new AI paradigm, so you can use the agentic systems
- 22:58Speaker 2: to bring older vertical sectors into that. So that's one
- 23:04Speaker 2: interesting thing that's happening. The other interesting thing is happening
- 23:08Speaker 2: is that research in science and technology and medical it's
- 23:15Speaker 2: accelerating at a pace which we have never seen before,
- 23:19Speaker 2: and that's due to AI. What you have is you
- 23:22Speaker 2: can do I mean this, you know alpha fold as
- 23:25Speaker 2: you will know. It's enabled people to do what they
- 23:30Speaker 2: would have done in ten years, it has enabled them
- 23:33Speaker 2: to do it in one day. So the pace of
- 23:38Speaker 2: research in biotech, in physics, in all the basic sciences
- 23:44Speaker 2: and bringing out new molecules, new materials, that's going to
- 23:49Speaker 2: become very very accelerated. And I think that's where we
- 23:53Speaker 2: are seeing a lot of traction as well. So that's
- 23:56Speaker 2: you know, so medical for sure, biomedical because we all
- 24:00Speaker 2: want to live longer and you want to find solutions
- 24:04Speaker 2: to it, and AI helps there. But you know, we
- 24:07Speaker 2: have to be clear, these are different types of AIS.
- 24:10Speaker 2: They're not just one big which I think is what
- 24:13Speaker 2: people think AI is one and there's just everything is
- 24:16Speaker 2: in it. One large language model can solve everybody's problems.
- 24:21Speaker 3: Yeah, that's a great point.
- 24:22Speaker 1: And recently we saw you know, Trump had all the
- 24:27Speaker 1: tech CEOs there and they were announcing billions and billions
- 24:30Speaker 1: of dollars of investment for the medical based AI and
- 24:35Speaker 1: like care and cancer, all these diseases and you know, immortality.
- 24:39Speaker 3: Who knows with AI. So it'll be interesting to.
- 24:42Speaker 1: See outside of finance and blockchain that I'm sort of
- 24:47Speaker 1: stuck in the medical side and how that's going to
- 24:50Speaker 1: advance over just the coming years.
- 24:53Speaker 2: Yeah, I mean that's that's to me, that's very exciting
- 24:57Speaker 2: because that's where you're going to see a lot of action.
- 25:00Speaker 2: That's where you're going to see a lot of people
- 25:02Speaker 2: doing interesting research, interesting stuff, and that's yeah, that's a
- 25:10Speaker 2: space to watch for sure.
- 25:11Speaker 3: Definitely.
- 25:12Speaker 1: I want to touch briefly on the crypto and blockchain side.
- 25:16Speaker 1: You mentioned earlier about the agents needing to transact on
- 25:20Speaker 1: your behalf. Is there infrastructure with fetch right now for
- 25:25Speaker 1: them to have a wallet and pay with stable coins
- 25:29Speaker 1: and you know, interact with different chains.
- 25:31Speaker 3: How does that work?
- 25:33Speaker 2: All of that is, all of that is in place
- 25:36Speaker 2: everything is actually we're not talking about it, we're not
- 25:39Speaker 2: writing papers about it. We are actually showing it. It's
- 25:42Speaker 2: available right now. So when you go to our own
- 25:47Speaker 2: large language model interface, which is called a SI one
- 25:51Speaker 2: dot AI, you can actually interact with these agents right now.
- 25:56Speaker 2: You can build your own agent. Within a couple of clicks,
- 26:00Speaker 2: you can actually deploy it and you can actually speak
- 26:03Speaker 2: to it. Other people can communicate with it. You can
- 26:06Speaker 2: communicate with other agents. So for example, agents representing machine learning,
- 26:10Speaker 2: other machine learning models. They could be representing any of
- 26:13Speaker 2: the hugging face models. You can actually ask it to
- 26:16Speaker 2: do protein analysis. You can ask it to do medical
- 26:21Speaker 2: analysis on a scan. You can ask it to generate
- 26:23Speaker 2: and report. You can do all of those things using agents.
- 26:27Speaker 2: Or you can say, hey, connect me to Ashton. I
- 26:29Speaker 2: could say connecting to as. If you have an agent,
- 26:32Speaker 2: it'll find it, it will connect me to you. And
- 26:35Speaker 2: you know, our agents can do whatever they need to do.
- 26:39Speaker 2: But the value exchange, the economic value exchange, is very important.
- 26:42Speaker 2: So all these agents have their wallets, they can make
- 26:45Speaker 2: payments to each other. They can actually have payments in ESCRO.
- 26:50Speaker 2: We're building trust mechanisms so that people people can trust
- 26:54Speaker 2: who the agent they're communicating with, is it real or
- 26:57Speaker 2: not real? So for example, if if you have a website,
- 27:01Speaker 2: then the agent also gets the trust level from the website.
- 27:03Speaker 2: If people believe in your website, they will also believe
- 27:06Speaker 2: in your agent. So that's kind of all the trust
- 27:09Speaker 2: mechanism that they're not They're not the future, they are
- 27:12Speaker 2: the now. So it's happening.
- 27:14Speaker 3: It's very cool.
- 27:15Speaker 1: And with the you met you mentioned like building some
- 27:20Speaker 1: trust mechanisms, I'm sure there's so much more to be built.
- 27:23Speaker 1: What's part of the focuses for the next twelve months
- 27:26Speaker 1: in growing ASI one even more.
- 27:31Speaker 2: Well, our focus really is to get this technology in
- 27:35Speaker 2: the hands of everybody. Everybody doesn't have focused enterprise or anything.
- 27:40Speaker 2: We want this. As I was saying, when you go
- 27:43Speaker 2: in a gathering of AI people, you say, hey, I
- 27:47Speaker 2: mean you've heard of agentic systems, yes, but do you
- 27:50Speaker 2: have an agent? Then ninety five percent of the people
- 27:53Speaker 2: actually in some cases nineteen nine percent of the people
- 27:56Speaker 2: have never built one or owned one. And that's a
- 28:00Speaker 2: change because you have to give thing to just do
- 28:03Speaker 2: it very easily, very quickly, and they have an agent
- 28:06Speaker 2: they can interact with it. So that's step one. So
- 28:08Speaker 2: my biggest focus at the moment is commercially in terms
- 28:11Speaker 2: of commercialization is what Chad GPT did, which is, let's
- 28:15Speaker 2: put the technology in the hands of people and let
- 28:18Speaker 2: them use it. Tell us what you want from it,
- 28:22Speaker 2: and then we will build that and that. But because
- 28:26Speaker 2: because the infrastructure is ready, and of course it's going
- 28:30Speaker 2: to keep improving and keep evolving and keep changing, but
- 28:33Speaker 2: the initial infrastructure is ready to deliver that to people,
- 28:37Speaker 2: which means that they can they can actually take an agent,
- 28:43Speaker 2: they can interact with the agent, it can ask to
- 28:45Speaker 2: do some tasks, and then then the whole evolution starts.
- 28:49Speaker 2: How how do you use the agent? How do you
- 28:51Speaker 2: use the AI behind the agent? What do you do
- 28:53Speaker 2: with it? You know, we know lms are good for
- 28:57Speaker 2: completing your your modules, your academic modules, you know, writing
- 29:02Speaker 2: papers and doing the textual things. What more can we
- 29:06Speaker 2: achieved because there's only so much you can do better.
- 29:08Speaker 2: So that's that's a push this next twelve months. We
- 29:12Speaker 2: want to onboard businesses. We want to onboard developers who
- 29:15Speaker 2: can build machine learning models. They should all have agents
- 29:19Speaker 2: being representing them and people should be using all of
- 29:23Speaker 2: that technology just from a simple interface.
- 29:27Speaker 3: It's very exciting. Hamiya.
- 29:28Speaker 1: I'm going to check it out for sure. I'm going
- 29:31Speaker 1: to start today as I one dot AI. I'm gonna
- 29:34Speaker 1: i'll put the links in the show notes as well.
- 29:37Speaker 1: And I think it's sort of the same for bitcoin.
- 29:39Speaker 1: You know, when people they've heard of it and but
- 29:43Speaker 1: and they're on the fence, but they're like there's something
- 29:45Speaker 1: stopping them, like why don't you just you know, get
- 29:47Speaker 1: two bucks, you know, like just try it, and that's
- 29:50Speaker 1: the only and once you get it, you understand the
- 29:52Speaker 1: value that it has. And we need to have that
- 29:55Speaker 1: for AI agents as well.
- 29:57Speaker 2: Yeah. Absolutely, But it's coming.
- 30:00Speaker 1: And I know FETCH and ASI are doing a great
- 30:02Speaker 1: part in that. I love what you guys are working on.
- 30:05Speaker 1: Please keep up the great work. I would love to
- 30:08Speaker 1: have you guys back on in the near future to
- 30:10Speaker 1: hear about more updates and how Agenic AI and ASI
- 30:14Speaker 1: is growing.
- 30:15Speaker 3: And thank you so much for the time today.
- 30:17Speaker 2: Appreciate it. Thank you Ashton. Good to be here.