The Wolf Of All Streets - In the Matrix with Humayun Sheikh, CEO of Fetch.Ai

Episode Date: September 3, 2020

Humayun Sheikh is the CEO of Fetch.Ai, a company solving today’s challenges surrounding blockchain and artificial intelligence. Previously a professional commodities trader, Humayun leveraged his engineering background with his interest in blockchain to improve the way machines interact with each other. Humayun is now designing tomorrow's digital economy, creating what will look to many like a Hollywood Matrix-like reality.  Scott Melker and Humayun Sheikh further discuss artificial intelligence for dummies, combining commodity trading with blockchain development, the dangers of AI, how far we are from the Matrix, teaching machines to interact with each other, blockchain reducing the middleman, self-driving cars, DeFi’s innovation harming traditional banking, the use cases for blockchain outside of finance, DeepMind, and more. --- CHOICE IRA by KINGDOM TRUST Don’t be part of the 7.1M Bitcoiners who have bitcoin and a retirement account but don’t have bitcoin in their retirement account. With Choice IRA by Kingdom Trust you can hold bitcoin in your retirement account. The first 1,000 users to open a Choice IRA will receive $62.50 in free BTC - visit RetireWithChoice.com/WOLF to join the waitlist and secure free BTC. --- VOYAGER This episode is brought to you by Voyager, your new favorite crypto broker. Trade crypto fast and commission-free the easy way. Earn up to 6% interest on top coins with no lockups and no limits. Download the Voyager app and use code “SCOTT25” to get $25 in free Bitcoin when you create your account --- If you enjoyed this conversation, share it with your colleagues & friends, rate, review, and subscribe.This podcast is presented by BlockWorks Group. For exclusive content and events that provide insights into the crypto and blockchain space, visit them at: https://www.blockworksgroup.io

Today's episode is brought to you by Choice, by Kingdom Trust, and Voyager. We'll learn more about them later on in the episode. What's up, everybody? This is Scott Melker, and you're listening to the Wolf of All Streets podcast, where two times a week, we talk to your favorite personalities in crypto, Bitcoin, finance, art, music, sports, politics, and basically anyone else with a good story to tell. This podcast is powered by Blockworks, a media company with over 20 podcasts in their network. You can check them out at blockworksgroup.io. If you like the podcast and you follow me on Twitter, you should also check out my website at thewolfofallstreets.io.

That's where I share basically everything that comes out of my brain there and my newsletter, which you can sign up for on the website. So today's guest is an entrepreneur, investor, and visionary, best known for his expertise in blockchain and artificial intelligence. He's currently the CEO of Fetch.ai, his fourth major venue where he is developing tomorrow's digital economy. Humayun was also an early investor in DeepMind, a company that was famously later acquired by Google for $500 million. So Hum Humayun Sheikh, thank you so much for taking the time and welcome. It's great to have you. It's a pleasure to be here, Scott. It's wonderful. Thank you for inviting me. Okay, so doing due diligence for this show,

almost everything I read was way above my head. So, I feel like probably you get that a lot. We can change that. Yeah. So maybe we treat this like artificial intelligence for dummies or explaining it to a 10-year-old. But I would love to start, I guess, from the beginning, how you got into artificial intelligence, how you became interested in this space and how that's evolved for you. Sure. So my background is engineering. So by trade, I was an engineer, trained up as an engineer, computer science major.

And I got quite involved into building some machine learning algorithms to do some commodity trading, predicting markets, price predictions, and all that kind of stuff. But then I also got introduced to Demis, who was a really clever guy who was working in the machine learning AI side, and I got even more involved with that side of things. And so we worked together for three to four years. It was quite interesting to see, you know,

we were starting to move towards this new wave of how can we think like humans and, you know, how can we make the machines more human-like. So that's kind of my, you know, over the last 10 years, that's where I've been kind of involved in. But what was quite interesting is that my co-founder, Toby Simpson, and Thomas Hain, both of them, we worked together on nearly three to four years, trying to work out how we could actually build something which actually you could start commercializing, really. And commercializing in the sense that, you know, how do you bring it down to the level where you're not just relying on this big data, but you're trying to build this

kind of digital world, which is quite easy to cut. Well, it's not easy, but it's very doable in a MMOG type scenario. So you have these different bots running around. You have all these things kind of interacting with each other. But when you come to the real world, things change, although they're quite similar. The issue with the real world is they're controlled by different entities. And what is quite interesting is that the whole world is like a multi-agent system because you have different agents. Everything is an agent.

If you treat everything as an agent, every human, every entity, every organization, then you start seeing some complex interactive behaviors because, one, you can't always have the incentives aligned. So that's a big problem. And then you have a kind of integration problems, which is, you know, how do you put two things together? How do you speak to each other? You know, there's languages, there's the same goes for all the machines. So this has been kind of a pet problem we've been working on. What is quite interesting is along comes blockchain, which came for a different reason, really, for more financial reason. But if you look at it, what it is is a big multi-agent system because you have all these stakeholders who are kind of not always aligned.

They have different incentives. They want to do different things. So how do you bring all of them together? Which is exactly what is happening in the real world where you have all these devices which are owned by different things, different people, different entities, and they're generating all this data. So what are we doing with it? So that combined with trying to figure out a way where we could start looking at creating these agents,

which are semi-autonomous, and they start interacting with each other, and then they start learning from each other. So when I was in deep mind with Demis, Demis' objective was mainly AGI the general intelligence side which is that how can we make the machines think like humans but but I'm a little bit on a completely different scale which I'm thinking how can we make the first step towards collective intelligence so rather than trying to do it in one place with one entity, why can't we do small intelligences like a swarm behavior? How can we put them together? How can we make it collective? And how can they learn from each other? But the biggest problem is, where is the fabric

which enables them to learn from each other? There isn't one. So that's what Fetch is building. So interesting, because to probably your average listener, it sounds like we're in the matrix and you're talking about agents and, you know, we're all being chased around, but it seems like that's not far from reality in that regard. That is actually very true. This is exactly what it is, because what you're building is a fabric, a matrix where you can have different devices being represented by these software agents. And they kind of connect with each other to do different things. I mean, a very good example is today, if I want to get my phone to speak to my car, I just connect it to the car.

Now, suddenly, what if my phone needs to speak to the traffic lights? I mean, where's the network? Where's the architecture? And how can they do it safely? I don't want my data going anywhere. I mean, with all the technology, with all this huge attack surface, which is being developing and all the hacking which is happening. Data, people are hungry for data. But what I'm saying is let's convert that data into information and find a way so that each device, each entity, each stakeholder can actually exchange that data in a way that it's not actually raw data. It's more information based from that data in a way that it's not actually raw data it's more information

based from that data and then it can learn from each other so how do you how do you create that open framework where this can happen and so that is what you are doing of course you're trying to find a way for all these different things to interact which seems with so many people, you touched on this, but there's so many companies and so many developers and so many people working on different technologies. Aren't a lot of those going to inevitably can be conflicting? And that is exactly the same point, which is that, yes, they are all conflicting. but what you want them to do is people can build their own devices, people can build their own protocols, but you want a common protocol where they can actually interact with each other.

So kind of creating an intelligent connectivity, that's that really is... So at some point, even those disparate projects and stuff are all going to have to integrate to this one larger platform, which hopefully is fetch. And that's, that really is the objective. But what we're not trying to do is we're not trying to control it that, you know, we don't want to define all these APIs. The objective is to make it so open that people can actually, you know,

the agents can actually adapt to it rather than trying to, rather than trying to hard code it, it's more adaptable. So the agents can actually decide how they want to communicate with each other. So they can create their own interaction models. They can create their own interfaces. They can create their own language. So you touched on this before that obviously blockchain is, at least everybody, I think their natural inclination is to think about it with cryptocurrencies and to think about

it in a financial sense. And I know that that's probably a challenge for you in explaining why you use blockchain when that's not necessarily what you're doing. How do you leverage blockchain technology to do what you're doing? Yeah, that really is. You're absolutely right. I mean, I get this question so many times and, you know, everybody is always asking, well, you know, what's that got to do with cryptocurrency?

Cryptocurrency is a very small subset of blockchain. And as I said, what blockchain is doing is it's creating a multi-agent interaction platform. So what you have is you have multiple stakeholders, multiple entities who are trying to do, who have their own objectives, but they're still trying to go achieve the common goal. Somehow they come, you know, so you, you know, if you have the culprits that you have to kind of weed them out effectively because the incentive structures need to be right. So now if you think about devices, if you think about who's going to make money from those devices, what kind of data needs to go where, you have to have prediction systems. So, you know, what are you predicting things are going to do you need to find a mechanism of aligning all the incentives which is exactly what blockchain does

right yeah that's that's um that's that's the common problem with interacting with so you're basically making sure that everybody who's participating is incentivized to behave well. Yeah, absolutely. And that same applies for cryptocurrencies and same applies for the multi-agent system. Can you talk more like, I guess, brass tacks, like specifically where are we seeing implementation of this technology is it you know when I want to call an uber and my smartphone needs to figure out where I am and where the car is and how to make that transaction is it you mentioned traffic lights what you know what are the real on-the-ground use cases for this so so what I what I could, I mean, what I'm going to just try and explain is

one example, and then you can extrapolate from that several other examples, which I'll please. But if you look at a very simple, a taxi booking facility or a ride hailing facility. So in a ride hailing facility, what's happening is effectively you have an Uber-like entity which sits in the middle, and all they do is they give you the interface, and then all the stakeholders which need to connect, which could be the actual service provider, which is the taxi, the cab, or the user, which is the consumer. Now, you have all of these entities, and if you look at a top-level view, what's happening is that you are enabling them to connect with each other.

And for doing that, Uber charges, let's say, a X percent fee. Now, why this is such a big system in the middle is because it has to cope with all the consumers and all the service providers. Now, what if you gave every consumer and every service provider its own small software, the agent, autonomous economic agent. So, and then in that agent, you define exactly what you want the incentives to be and what you want the objectives to be. So you, you give them policies. So you're assigning them policies that, you know, if, if, if it's between this time, I don't want it. If it's between this many miles, I don't want to do it. Sorry. Are you saying you being me, the person who's hailing it,

or you as the programmer who's creating this system? This is nothing to do with the programmer. We provide a framework where you can put all that in there. And now you suddenly have this network where you have thousands or hundreds of thousands of agents, and you can communicate with them directly, negotiate with them directly, you can book with them directly, and you completely cut out this whole middle layer of intermediary service provider.

So all you need is you need to have an interface. Now, I'm not saying that there is no room for such a service as these ride-hailing companies are providing, because don't forget, there is identity issues, there's security, there's regulation, all of that is understood. But what needs to be looked at is a new form or a new way of dealing with this because as more and more entities interact with each other, you will see more complexity. And the complexity cannot be solved from a top-down approach. It has to be solved from a bottom-up approach. Because with the ride hailing, you ideally want to interact with the traffic light so that you know how long it will take for you to get somewhere. And, you know, the service provider or the consumer both need that information.

Then you want to speak to other cars to see, you know, what's the traffic. If they're closer, what the traffic situation is, right. Right. And all of these kind of issues, then, you know, now you have to try and deal with it in one big chunk as a big entity or as a big piece of software. It's very difficult. But if you enable the agents to make those decisions, to make those calls and understand each other, then the system becomes a lot simpler because you don't have to take care

of everything. You can interact between, one consumer can interact with five, six suppliers. It doesn't have to go through this central entity. And that's really one way. And if you now think about commercially, if you think about supply chain, for example, you have multiple stakeholders, multiple participants in the supply chain, which includes financing companies, which includes haulage companies, which includes lorry drivers who have their own scheduling to do.

Then you have the factory, which need to obviously do just-in-time productions. Then you have the whole delivery network, which you need to take care of. So now you're building a system which is quite a complex system. But that complexity can be reduced if you assign each stakeholder an agent with different policies and parameters, and then you let them interact with each other. And that really changes the game. And not only does it change the game, but it also changes the game financially

because now you don't have to rely on some middle entity to solve your problem. You can solve your problem yourself. Yeah, so is it a matter? middle entity to solve your problem. You can solve your problem yourself. Yeah. So is it a matter, I mean, obviously humans have managed a supply chain manually for hundreds of years to some degree, right? So this has been done without it. So is it a matter of efficiency and cost? Is that really the motivator here?

Is it so that things can arrive faster? Is it so that the providers can make more money and cut out some of the middlemen? Or is it something bigger? We're moving more and more. It's both, really, because we're moving more and more towards autonomy. Autonomous cars, autonomous. I mean, we want to be, for hundreds of years, humans have done the work. We're trying

the machines to do more of the work, right? So that we can have a better life in terms of more free time, more efficient, do jobs efficiently. So if that is the model we want to get to, what really needs to happen is that more machines need to interact with more machines. You need more data, you need to be more efficient. And if all of these things are going to sit in different kind of silos, then it's very difficult. But if you look at the kind of things which is happening where

companies have done really well, for example, Amazon. I mean, Amazon doesn't do one thing. Amazon do many things. And that's the complexity which I'm alluding to. Because if you're not that complexly made, then it's difficult. It will become more difficult for smaller companies to survive because it just, you, to be efficient, you have to be that size. And I think this is kind of a complete reverse approach,

which is we don't have to do it that way. There is a way where you could, and I'm not saying, there's no judgment here if this is the right way or the wrong way. Right, of course. I feel that is my preferred way. So, you know, I talk about this all the time, but with blockchain specifically, obviously, like we've gotten to a point with the internet where nobody questions how the internet works. They just use the internet. We've gotten to a place with our phones where you don't question the technology behind the phone. You just know that you have a smartphone and it works. How much is artificial

intelligence already integrated into the things we use and into our everyday because people probably just have no idea? Or how much is this really the future and we're just at the beginning? I don't think we are there yet because I think we're at the beginning of it. So let's define artificial intelligence. Artificial intelligence, in my opinion, doesn't mean, or I would say AGI, which is true artificial intelligence, which is when things can do things like humans and they become more intelligent. What we're doing is more cases where we're doing data crunching and we're doing machine learning. Now, that's, I mean, if you look at what machine learning is really achieving today, it's nothing that dramatic.

I mean, we were doing, you know, these are 1950s, 1960s technology. Of course, we've improved them considerably, and our compute power has gone up considerably. With quantum computing coming, perhaps even more will change. But the approach is still the same. Let's collect this big data. Let's process the data. Let's just number crunch it and let's then come up with some different strategies and learn from that. That's okay. But the problem is that that puts the data in the hands of a few because if you don't

have a system which can learn from small amounts of data, and we're all going on about big data, and there's a whole race, then you have this problem. I mean, how much data is enough? Because we're generating more and more data every second. So if we can't control our data, there will come a time we will just be passengers and we won't do anything. Nothing will be in our control. Somebody else will be controlling it. Now, this is

truly sci-fi sounding, but... It doesn't really. I mean, yes, but no. I mean, all these big companies are really data companies more than they are anything else, right? We watch what they want us to watch. They make before we know what we want to watch.

They know what we want to watch. The delivery is on your doorstep before you actually order it. So, you know, it's predictive analytics. I'm not saying it's a bad thing. It's obviously facilitating quite a lot of convenience. But it doesn't apply to everything. So, you know, if I go and sit in the car, I mean, I want to drive. If I'm going to enable autonomous driving, a lot needs to change.

And a lot needs to change, and a lot can change, but not with the current model. Because if you start putting, so who would control the traffic? Which central entity would control it would it be toyota would it be tesla would it be which one um and are they going to be able to cooperate like with each other i mean who's going to give data to whom i mean it's it's one of the most difficult industries right and uh new um kind of innovative ways need to come in. I mean, we did an example of a sign demo where we just assigned all the agents on the road onto all the signs. Now, you know, they're in nobody's control.

They're entities in their own right. And, you know, the signs, you know, somebody could just say, okay, I want to take this sign and I'm going to deploy a dynamic signage here. And the agent actually looks after and interacts with anybody. So it doesn't matter which car you are. You could give information, not data, information for decision making. And that structure needs to come in place. And I'm not saying it's an easy task. The problem is it doesn't exist. Until it exists, there will be no uptake. Right. So it sounds like there's a major risk of the data being held by a powerful few. So is partially what you're doing trying to democratize that data and

spread it out so that it's not centralized in that manner? Yes. That would be an outcome of it. We're not specifically saying, let's just go and democratize data. That's not our objective. There are other projects who have that objective. And I think that's great, but that is is not our objective our objective is that we want devices and entities to have the ability when i say ability it means the knowledge the information and the the decision making power decision-making power to actually learn and make decisions. And in doing so, what you would see is that the data democratization happens automatically. Does machine learning inherently, I guess,

does it drive better decision-making than humans? Yes, it can, because we see that all the time. We don't know things which machines could know. I'm not just talking about personal data, right? I mean, even a Tesla driving on autopilot theoretically drives much better than any human, right? And it can take a decision way before a human would do because it can. But now just there is this example, which is exactly kind of positions us.

So if you look at Tesla, so Tesla, for example, today would be looking at the visuals, right? So you have cameras, you have radars, and as long as those things are well within range, it will spot it way before humans. But what if you can't see a car? You know, 10 cars ahead, you can't see it. Now, what we're trying to do is that if there is an agent in that car, that car could communicate very easily because it knows, you know, which cars to send this information to. And if there is going to be a failure on the brakes, you could know without even seeing

the car. And that's really where the agents come in more. They become more powerful because now you have this new sixth sense, which is your agent, which knows, is saying, okay, anybody here which I need to communicate with, as soon as you find that entity, you communicate with them. And if there's a red flag, you get told about the red flag. You don't actually have to see it. Right. And you personally don't even have to do anything, right? I mean, your car is going to react on your behalf.

Yeah. So it's just the next step of improvement. So we're not just relying on the visual, you're also relying on this network of agents who are feeding information. It's funny, my daughter, she's five. And she asked me recently, I was driving her somewhere and she said, you know, when am I going to learn to drive? I said, you know, in Florida, it'll be when you're 16, but I don't think you'll ever drive a car. She's like, what do you mean? I won't drive a car. I was like, I really don't think that, you know, 11 years from now, people are going to be driving cars. Do you think that I was lying to my child?

Do you think that that's true? I think that's very true. I mean, unless you really wanted to and you, you know, you have something for cars to drive the engines. I think that's very likely that you're absolutely right. I mean, it's just really crazy to think about. So obviously, you know, artificial intelligence, this whole space has been widely dramatized in Hollywood. And so I think that that obviously has an effect on how your average person thinks about it. And there maybe is this fear. Are there actual threats like in the movies that this poses for the future? I mean, it's great that your car drives automatically, but you see movies like

Minority Report. What if your car just drives you to jail because it thinks you committed a crime when you get in it? It seems far-fetched, but. But, but you, it's, you know, there's always two sides to the coin, right? So you are going to have goods and bads. And, and, and I think, again, you know, I mean, I'm not going too much into Terminator world or sci-fi world, but there is a danger that that could happen. And the danger is also, I mean, what I'm more kind of thinking of referring is that a more immediate danger is that this AI is not inclusive. It's a non-inclusive AI. I mean, where is the data coming from? How are the machine learning algorithms running? I mean, do we consider what's happening in Africa at all

when we're building this AI? Or, you know, what is the inclusion? Are we taking it in the right way about the two sexes? Or, I mean, you know, is everything coming in? And that's more of my concern, which is that, you know, it could be, you know, the AI could grow into doing predictions which are not neutral. They could be, you know, influenced by what data you put in. And that is not going to change unless you enable all the stakeholders to add to that data pool or unless you enable them.

So, I don't want my agent to make a decision based on, you know, somebody living perhaps in LA because I don't live in la and my decisions need to be different and you know that's the inclusion which i mean that's that's a very um you know simple example but you know there could be more serious consequences of non-inclusion that's really interesting so obviously you um are tokenized you You guys have FET. That's how I've always called it, or F-E-T. I don't know, but I've always called it FET. I've traded it quite frequently on Binance without ever, admittedly, digging deeply into the use case of what FET is. So why did you decide to tokenize? What's the advantage of doing that? And I guess, what are the implications of having your currency being traded on the open

market? I don't see it as a currency, to be fair. What I see it as is we needed to have something in our system. So I'll give you an example. So you have an agent, You want to run an agent. There's a cost associated with running an agent. It doesn't matter where you run it. How does that agent run and what pays for it?

If you're in blockchain, I mean, if there is a solution which is already there, which is cryptocurrency, then why reinvent anything? Yeah, no reason to reinvent the wheel. There's nothing to reinvent, which is, you know, you have to have an economic value exchange at every step. Now, what you do trading, you know, that's something else. That's just the financial market, which I did say clearly, there's always going to be that implication because that's what crypto is all about. But what we really have is, so let's

say we're building this system right now amongst 20 hospitals where you have this, everything has a health data and the agents learn from each other to kind of look at, you know, lungs, x-ray of lungs to kind of do detection for either COVID or just pneumonia. And, you know, the couple of hospitals are in Asia, a couple in Africa, and then UK, US. And so all these things are now training each other. If they're training each other, what's the mode of economic exchange? Because you're not going to go and say, well, I'd like your credit card details. So you need a mean of value exchange. And FED is that mean of value exchange. You can build several other layers of value exchange on top of it,

but to run the system, to use the services of the system, you need to have a value exchange. And that's really what Fats' purpose was. Don't be a part of the 7.1 million Bitcoiners in the United States who have Bitcoin and a retirement account, but don't have Bitcoin in their retirement account. Seriously, you can hold Bitcoin in your retirement account and not just GBTC. How can you do this? Through a self-directed Choice IRA by Kingdom Trust. The first thousand users to open a Choice IRA will receive $62.50 Visit retirewithchoice.com slash wolf. That's R-E-T-I-R-E-W-I-T-H-C-H-O-I-C-E dot C-O-M slash W-O-L-F. Podcast listeners receive extra points to move up the waitlist and get their choice IRA first. Do it right now. It's time to take control of your financial future and free yourself from the restrictions of classic retirement accounts.

Are you sick of paying ridiculous fees to trade crypto? It's time you try Voyager. It's hands down my favorite place to buy and trade crypto, and it's 100% commission free. Voyager gives you easy access to more than 30 top crypto assets, and you can instantly transfer cash from your bank account so you never miss a trading opportunity. Even better, you can now automatically earn interest on your crypto holdings. Currently, they're offering 5% interest on Bitcoin and 6% on USDC. Yes, you heard that correctly, 6%. And there are no limits or lockups, which means your funds always stay liquid. Find out why so many people are making the switch to Voyager. Visit investvoyager.com or search for Voyager on the iTunes or Google Play store and get

$25 in free Bitcoin when you use the promo code SCOTT25. That's investvoyager.com, promo code SCOTT25 for $25 in free Bitcoin and start trading today. Right. So how does the volatility on the market and obviously the financial interplay with Bitcoin and all that affect the use case of FET? Because obviously it's not a stable coin, right? I mean, the value fluctuates based on the market. So how does that affect the actual use case when somebody needs to use it? I mean, just give a very simple example is that you take the volatility into account, but you then convert that into a per cycle, per instruction kind of cost. And if you fix that, it doesn't matter what the volatility is.

So it's fixed within the system, just not relative to other. So that's probably the easiest way I can explain it. But having said that, I mean, the whole space is evolving and better solutions might come along. As I said, we're not trying to reinvent that. Our focus is on what we do. And if a better solution comes along, we will adapt that solution. So obviously, I mentioned in the intro DeepMind, which was acquired, I believe, in 2014.

Is that correct? Yeah. And from reading about it, it seemed like it was sort of there were two opposing camps on how that should happen. And probably the reason that you're at Fetch now is because you're on one side of that. What was it like to be at DeepMind and to see these big players who obviously can use that data in any way they see fit coming in to acquire the technology and the company? Well, there were a couple of things. So my involvement was more commercial. So what was quite interesting is that

these solutions are very difficult to commercialize, right? So for example, you can play the game of Go, great. But if you're a small startup, how do you fund it? Because you have no commercialization strategy, right? So it was very obvious that it was going to be taken over by a big player who can who have deep pockets and can do all of that stuff right now um and that is you know so for example google has huge amount of compute power you know you could run huge amount of compute power to win a game to do anything. You can do number crunching. But that's not really the whole objective here.

Because if that is how we're going to be, then we're going to build this big central AI somewhere, which is going to just control everything. So who controls that central entity? Yeah, that's the matrix. And that's the matrix. And that's the problem. I don't know if you've, I think it's Life 3.0. I don't know if you read this, but Max Tagmark wrote that, which is, again, you're building more and more value

within one company, which is controlling everything. So that was my point of deviation there. It's for everybody. It's not for one company. And I think what I would say is that if more money was available, and I think UK is probably not the best place to get this kind of blue sky thinking money. I think DeepMind would have done much bigger things and it would have sold for

probably 10 times what it is. So I think they're building something amazing and I think, you know, but for the time it was the right decision because to raise more money without a commercialization strategy, and they haven't still commercialized anything, to be fair. If you look at the commercialization, the benefit is always going to Google. So you can save some energy, yeah, but for one company's benefit, right? Right, right. Not for the benefit of the whole world. So we've talked about data and I guess the risks of it being controlled by a fortunate few.

Do you view these big companies as the bigger threat or centralized governments? I mean, who really wants to control this data and is incentivized to control this data and can use it, I guess, for their own interests and not the interest of your average person. I mean, I have a more philosophical thought on that. If you start from the beginning, we had, you know, religion was controlling things. Then, you know, the hypocrisy was controlling things. Along came democracy and the politicians were starting to control it. I don't think governments are controlling thing. Along came democracy and the politicians were starting to control it. I don't think governments are controlling anything. I think big corporates are controlling

right now. And as we go forward, that's going to happen more. And the reason is because the governments just are so far behind. They, you know, they just can't, I mean, they can use force to control it for a while, but that's not going to continue being the controlling power. Because what big corporations, technology companies have is kind of becoming unstoppable. So what the government can do is it can actually enable different mechanisms which break this big corporate structure. And I don't think they're doing the job right because they're not actually investing in that kind of approach because there is always this, you know, If VCs are going to invest, VCs are always going to invest in companies which are going to make money. Then the incentive is to make that one company bigger. Again, coming back to the multi-agent system and the blockchain argument, you are not going to get the same result if your incentives are so misaligned. They're hugely misaligned.

How do you make money? Because you don't make money by losing control. You make money by gaining control. So you are going to try and get control. And who do you make money from? Somebody is going to pay you that money. So what you're doing is you're taking it from individuals and that's how the companies make money. Yes, some companies add value, but it's my loss is their gain. Right. So it's interesting. We saw Facebook at least attempt and who knows where that project

stands, but attempt to make you move into money with Libra. So now we're already talking about the dangers of these large corporations, and now they obviously want to control the money, which then brings us into the bigger topic of central bank digital currencies, which we see are being tested. And we know that money is going to all be digital very soon, right? Which is more data, less privacy, more issues for the individual. I mean, what do you think of these companies and then governments, you know, moving into the digital money space? I think there's always going to be this polarization of the two. So you can't really stop that. And with money will come power.

But I think my point is, unless these governments have that much technological lead, they will not be able to catch up with right they're always behind i guess i'm just talking more about the risk of you know at least now we still have cash right i can make a transaction with you and nobody knows if all currencies become digital whether run by controlled by a company or a government every single transaction will be traceable every single transaction will be tracked there will be literally no way to transact privately. Yeah. And, and I think, I think we just, I mean, this is a personal, I think it's coming.

If we're going to go with the centralized approach, it's coming. If we don't, then, then cryptocurrency is, you know, is, is becoming, yeah, that is the only way, I guess. It's a rebellion. It will be the only way to operate outside that system. But that doesn't mean if it's good or bad, because you have the negatives of Bitcoin and cryptocurrency as well. I mean, you know.

What are those negatives? Well, I mean, the dark web is not like hidden from anybody, right? And then the thefts which can happen on digital currency, there's nobody to monitor it. Once you've gone, it's gone. So we're kind of going back a step to come back again to the same place. I mean, this was happening when there was no central regulations, and that's why the central regulation came.

Now, it's a delicate balance. I don't think we have an easy answer here. I mean, I like to think that it's a positive, a net positive for Bitcoin and cryptocurrency for those of us who believe in it, because, well, I guess for a few various reasons, A, being that this grand awakening that you have no privacy and maybe you would use it to increase that. But also just, A being that this grand awakening that you have no privacy and maybe you would use it to increase that. But also just, I think the general idea that if currencies become digital, people will become familiar with transacting digitally. They'll understand how to use a wallet though, you know, and,

and those things. But I wonder, you know, what the end game is there with, with these centralized entities, controlling the money even more. So that's why I was so curious. Well, I mean, again, DeFi space is doing a wonderful job here. I'm not saying everything is going to stick, but I think there is a huge amount of innovation which is coming, which will definitely affect the way we operate.

We're already seeing that. I mean, again, I'm looking at my background is in commodity trading, as I said. I'm looking at all these big banks pulling out of funding commodity trades. And, you know, there's no financial incentive for them. Now, if you think about it, and that's one of our solutions which we are building on top of Fetch, if you think about it, the new wave of these DeFi, like the compounds or the RVs and liquidity pools, they're actually quite interesting innovations. And I think if we tailor them right, I'm not saying central banks are going to disappear or go anywhere soon and the governments are going to stay.

But certainly this is a big threat to big banks. That's for sure. And I'm glad that it is because I think. Of course. It's interesting. Now we're seeing at least we've seen movement in the United States in that direction from the banks. We've seen, you know, that they will now theoretically be allowed to custody cryptocurrencies, which I think is huge, which then means that banks will be able to hold your Bitcoin and lend digital or fiat, you know, or dollars, correct? And then you're going to see a lot of assets, the physical assets being converted and tokenized and converted into digital assets. So that phase is coming. And if that phase comes and then you have these autonomous systems where you can lock up liquidity,

you could take your physical asset, convert that into liquidity and borrow against it, then banks become a little bit redundant. And I think, you know, this will give the banks a run for their money. It's interesting, it's all about lending and it's all about leverage, right? I mean, when you think of it, when you kind of zoom out and think about that, it just, it's a little scary. I mean, and again, you know, with all this, what's been happening with COVID, a lot of banks are pulling out of lending at the moment. This is the time when we need more lending. If the economy, or borrowing, if the economy has to kick start. And again, you know, who's to say this won't happen again?

And I think there will be other financial things happening. If we've learned anything in the last 12 to 13 years, I think we can pretty much be sure it will happen again. So, yeah. So again, so that's, that's, that's really it. And I think, I think it's, I mean, really excited about this space at the moment and, and, and how, how we're developing it in terms of different angles. But again, my point is financial, yes, it's very important,

but I think there's more to it than financial because if you look at the physical things happening on the edges, they need to be integrated with this new financial architecture to actually have an impact. Because you can't just live in this little bubble. The bubble has to start interacting with the physical. So how far away are you at this point from making all of that a reality to the fabric that we've spoken about, to having that completely built and ready to go

and have basically everything interacting with it? At the moment, the deployments are starting. We are deploying some trial supply chain solutions. We're trialing with small service providers for ride hailing. We're building a network of, we're calling it DDN, which is a kind of decentralized delivery network where you can actually connect ride hailing, parcel delivery, food delivery into one. Which means that you give the control back to the

service provider and the consumer, and you let them interact with each other. And within that system, then you can keep adding more features because, to be fair, it's very easy to add feature because you have an agent. You're not defining anything in the middle as such you're letting them communicate so what that does is you can add different features so for example from the DDN moving on to hospitality where you can actually you can actually book hotels rooms housesals, it's a very small step. So you could actually enable kind of individuals to do their own thing with the confidence and the comfort of having a trusted entity in the middle because you're kind of giving trust to all these agents you're

giving them decentralized ids where you can actually transact with them knowing that it's okay to interact with them so that's the next step in that evolution yeah i mean we went from you know you call your travel agent to i remember being mind blown when like travelocity and expedia became you know so uh universally used. And then we go to Airbnb, which is like, it's not a hotel anymore. It's someone's property. And now we're eventually eliminating the middleman there as well. So you're transacting directly with the person who owns that house or the hotel or whatever. Because that is exactly what you do today. I mean, if you take out the secondary things, that's all you do.

When I book a hotel, I book a hotel with the hotel chain. The only thing is because technology says, okay, I'm going to be able to aggregate all of that for you. But what if you don't need aggregation? What if your agent could query everything directly? You don't need aggregation. And then you can actually negotiate with hotels as well. I mean, your agent could be saying, okay, I've got three hotels. It's the same vicinity.

I know it's the right place. I can negotiate. And one of the hotels has full occupancy, doesn't want to negotiate. The other one might. So how do you find the best deal? Not the best deal for the middleman, but the best deal for you. The best deal for you as the consumer. It's interesting. It's like when you talk about these agents, it almost sounds like every individual is going to have access to a high-powered personal assistant.

And that's exactly what we are hoping will happen over time. That is coming. I mean, we can see that with Alexa. We can see that with Siri. You know, they're helping. The problem is still the same, though, because Alexa's objective is not aligned with your objective. Alexa wants to provide data to Amazon so that they can sell me things.

So, you know, so that's really, I mean, I'm not saying it's always that bad, but I think there's too much of it. So it would be a personalized Alexa. What are your thoughts on things like Alexa and Siri? And should the average person be using them? Or is that a very, you know, personal decision based on your own comfort level with privacy in these companies? I love technology.

I always like to use it. It disappoints me many a times. But that's not to say it's not good. I mean, the other day, because there was quite hot weather in the UK, you know, the temperature kind of went up quite a bit. And Alexa comes and says, well, you know, if you're going out, you be careful. It's quite hot. Now this was unprompted and it's quite nice. I like it. I like it. But other people might think it's a bit creepy. Yeah. I think that in the peak of COVID, Alexa became my daughter's best friend.

So I can appreciate play games with Alexa. She plays whatever music. I mean, I think it's amazing that she can ask it questions. You know, we have help that primarily speaks Spanish, and she translates things and talks to her, you know, by asking Alexa. I mean, it really is pretty incredible. And that will evolve from there so that you can just say, well, Alexa, book me a cab.

And you can do that, but it's not structured correct. That's why there's so much hindrance because there's so much friction because it's not structured in the right way. Now, if you imagine Alexa connecting to your agent and the ability to connect to anything else, then Alexa would be much more efficient because you could say, Alexa, go and book me a cab. Right. Um, and Alexa could go and query all the cabs and, you know, negotiate the deal and you have the cap or, you know, or book a hotel like that. So rather than going through 10 other loops. Right. So, I mean, at the end of the day, we're just trying to eliminate middlemen from everything. Not from everything, but trying to do it where it's more efficient to do it without them.

Because there are more efficient ways. I mean, forget about the middleman. I mean, I've got nothing against the middleman. You know, I'm a commodity trader. I mean, I'm the middleman. I mean, I've got nothing against the middleman. You know, I'm a commodity trader. I mean, I'm the middleman. That's not the point. The point is it's not efficient anymore because you could do it more efficiently with a lot less cost. And I think that's really the key here.

It's really, I mean, it's just amazing to me because so much of it happens in the background and I don't understand it. And hearing you articulate it now really does give me a greater understanding of why it's so important and what's inevitably coming. There's no stopping this, right? I mean, this is the future. This is definitely. also this is this is one of the reasons why we feel quite bullish about fetch as well because if you look around in the crypto space there is nobody else building this i was gonna ask if you had any competition there is nobody doing right so we we're saying you know you know it's like, okay, so, I mean, just to give you an example, you know, the service like Chainlink provides this Oracle-based service, right? Great service.

And what it is, is effectively just an agent, right? So you're getting this data, although it's authenticated data, when you need it, it's providing it to you. Now, try and do this, not just with prices of cryptocurrencies you now now try and do this not just with prices of cryptocurrencies now try and do this everything now your taxi um needs a comparison you know what's what's the rates what's the ongoing rate you know how do you decide i mean uber can charge you know 2x um you know 2x fee extra fee because congestion, right? How do you do it? Right.

So do you have to be integrated with an oracle like that, or is that part of Fetch's value proposition? Fetch has got its own agent-based oracles, not for pricing, not for what chain link does. Having said that, it's not that difficult to do that either, but that's not our core. Our core is to provide oracles for agents and they can make decisions. So, for example, let's say you're training your algorithms, you're training your system to detect something.

Let's say pneumonia, and you have various hospitals participating in it. So you need to see what kind of rates, what kind of levels of prices, what, who needs to be rewarded, how much those oracles are always needed. And that's one of the reasons why we need to build our own when we can. Right. I mean, how much is there risk? I mean, obviously, they're verifying information is true for anyone who doesn't understand at the very basic level. You know, how much risk is there of incorrect data impacting the decision of an agent? Or, I mean... There is, of course, there is, because if the data is not right and there is no mechanism to verify that data,

then obviously you will be making wrong decisions. And that's exactly my point. I mean, people expect we should be commercializing in two minutes. It's not a small problem. It's a complex problem with different various components which we need to take care of. And again, if, for example, Chainlink starts providing us taxi data or weather data, which it could, we will take it from them. We don't need to reinvent anything. Right, of course.

Our core is the agent framework. Our core is to build a framework where you can run agents, where even if they are, for example, financial systems, they are agent-based financial systems. And that's really what we're trying to build. But again, Oracle play a very important part in it. I read a quote that you said, and I don't have it in front of me, but it was something about in a market,

making predictions is as easy as finding a dinosaur in a garden or something to that effect. I'm wishing I had written it down, but I remember kind of reading it. What did you mean by that? I can't remember that. But, yeah, I think the focus here is to have prediction markets. And when I say prediction markets, it's not just financial prediction markets. It's more prediction markets which enable these agents to do something.

So what is the prediction? So just a very simple example, if the weather is 90% is going to rain, then the agent suggests like Siri does, take an umbrella, right? Yeah. Correctly. So agents can't do anything without predictions because you need them now again learning from the financial markets again which is like the projects like auger does the prediction market but they do it for a different purpose we're doing it for a

different purpose and it's not it's not structured the same either but you need to create that incentive so that if I need an information from outside to make that prediction or I need to kind of make a prediction, I need the incentives aligned so that I get the right prediction to make an action. So it's, you know, actions are based on predictions. I mean, we do that all the time. I mean, if you're going to buy a token, your expectation or your prediction is going to go up. Right. And so as a trader, and you obviously were a trader as well, how much do you think that, you know, quant obviously and all these things are, you know, crunching social data, things like that. How much do you think that

we can actually predict price movement of anything? I mean commodities, which we're focusing on, and we've just announced the launch of a tokenized decentralized commodity exchange. I think it's a huge amount. So if you look at agricultural commodities, if you look at, you know, you could certainly make very good predictions there. Less so in crypto, because who knows? Some of these things you don't know. But having said that, I mean, sentiment plays a very important role.

So I think predictions are very important in making all sorts of decisions. And we do that without knowing, right? Yeah, I've always traded crypto markets primarily on technicals because it's the only market where fundamentals kind of fly out the window and you can actually look at a chart and kind of obviously assume what people are thinking, which is all you're really doing right. But I have seen some very compelling predictive modeling based on social, you know, based on Twitter, and it's becoming very compelling to me. Yes, and I think you're absolutely right. And it applies to, I think it applies to commodities.

And I mean, it also applies to, for example, housing. It applies to, you know, everything. I mean, so, and this is again, my point, which is, you know, you have all this different data which you need to pull in. And then some of these social parameters and some of these social indicators you might not find on Twitter, because, you know, not everybody's going to talk, for example, about housing. Right. It works in cryptocurrency because that's where the community is and sentiment is very easy

to get. I mean, naked eye, you can gauge sentiment. So when you add a tool to that, it becomes very compelling, I think. Yeah. So I think we will see more and more of it because the new generation is only speaking on social media, right? They're starting to lose the skill to physically interact and speak. I believe it's lost, sir. From what I've seen, I don't know how old you are. I'm 43, but it's just crazy. I mean, it's crazy how much that has changed since we were kids.

So I know that we're up against it with time here. I would love for you to tell us where everybody can follow your progress, keep up with the project, follow you individually, and make sure that we always know what's happening with Fetch. Yeah, so for developers, we have a developer Slack channel. So please join it. We now have active, we're trying to actively encourage developers to join

because the agent framework is ready. You could deploy agents, you can run agents, you can make different applications very quickly. More out-of-the-box solutions are coming in the next few months. So keep an eye on that. We have a Telegram channel where you can actually have, yeah, I mean, I think you can ask us questions. We do AMAs.

You can ask any commercial type of questions. We also have all visibility on our website. We kind of show different crowdcast interviews to show what we've been building. We do regular crowdcasts to show all the examples of what we have been building. We encourage people to have a look. It's a very open framework, and there is no specific application which this applies to. Because machine learning and AI ultimately will have to become a commodity. So you don't have a choice. You have to use it if you're going to do anything.

And what we are trying to enable is everybody to be able to use it. So it doesn't matter if you're not a developer, we're building tools for people who can just... Everyday people. Yeah. Makes sense. So I have to ask one more question, even though I was about to finish, because it just popped into my mind, because I always laugh like Elon Musk saying that we're in a simulation. Do you think that we're all in one big simulation? I seem to agree with it. It's a biological simulation, but what does that mean? Nothing. It is a simulation. I guess what is a simulation?

It is still the same. It's still the same world. but i i do i agree i i think i i would definitely agree with that we are in a simulation and doesn't that imply that there's somebody controlling the simulation well not necessarily because an entity of some sort um not necessarily but it could be, I guess, again, it's how do you define entity? Because if you're just, you know, as you look at in these massively multiplayer online games, or as Toby, Toby was my co-founder, could spend days telling you this, you could just give very basic characters to some agents, and they can evolve. So not necessarily we don't have to look at doing something.

Now, if it's actually a really good true simulation, then not necessarily that entity or something had to exist. It could be randomness. I'm going to leave it there because now I have a lot to think about. Thank you so much for taking the time. I really appreciate it. I'm excited to see what comes out of Fetch. It really sounds like if this works, and as you said, it's sort of an inevitability that what you're doing will largely be the framework of our entire existence moving forward in the way that we interact.

Well, we hope so. It's better decentralized than centralized. You've convinced me of that if I didn't already believe it. Appreciate your time. It's been a very, very good conversation. Really enjoyed it. Thanks again.

Thank you.