A World of Human Possibilities I Mateo Sauton

Hello everyone, I'm Mateo, from TFH, a company that creates tools for the world. Today's seminar is going to be about... we called it "The called it "The World of Human Possibilities," because I'm going to talk to you about a few different products that we've released that we're looking forward to you testing and developing. I'll start by outlining where we are now as a company and what we're as a company and what we're likely looking for, and then I'll go over the SDKs we have and the directions for the hackathon. OK.

So the way I see it, and I've seen it, I've seen it, this, this whole ecosystem of technology in general, over the last year, is that everyone thought we wanted to prove that someone was a person on the internet, and we acted like we actually wanted to know, is there a person behind this internet interaction? And that's why we created one of the three SDKs that I'm going to show you today. This is called " Agent Set". Uh, and that's because practically the entire internet was filled with bots. Now the majority, somewhere around 45-50% of internet traffic, is made up of bots.

OK. So, AI is everywhere. We all know that maybe some, maybe some not. And we're trying to protect or help people interact online with other people, or at least know that something is interacting online on behalf of that person, without revealing any information about that person or people. Uh, but this is a really funny, uh but this is a really funny, uh , chart that I really like to use when I'm showing things like the complexity of a task and the time it takes a human to do it, not compared to what the US AI does.

Uh, so where we are now, we've just entered the third phase of the project. If you've read our white paper, we've talked about this probably a million times. Hmm, but we've gotten to the point where World ID is so good that businesses can use it reliably, and they're starting to use it, and we've gotten to the point where we're known enough around the world, for better or for worse, whatever, uh, that everybody knows what it is, or maybe a lot of people know what it is, at least a few million people, uh, and I mean, the plan is, we're going to , uh, let everyone on earth know who we are and try to join the network, uh, and become the largest, uh, financial network of people in the world. Uh, yeah people in the world. Uh, yeah , uh, and, and that gives us three different points, like, businesses, uh, know us, people know us, and of course, if those two know us, the agents also know who we are.

Uh, like I said, a few million people know about us. Worldlap currently has 38 million users. 18 million, uh, verified people worldwide and approximately 7.5 million active users inside Worldlap. Uh, this is a huge expansion, a huge growth for the project and for projects that use, uh, our technology.

You might not know, but there's one of the SDKs that we use, that we released, called the mini kit SDK, which allows you to create mini to create mini apps. Essentially, these are third-party websites inside Worldlap and these third-party applications using our distribution, as you can see, and it's been incredibly successful. I'm not going to talk about mini- talk about mini- apps and mini kits right now because we're going to focus on other things. But I encourage you to check it out, if you want, you can go to the documentation or to mini apps.orld and look at, uh, the entire distribution that mini apps have.

Uh mini apps have. Uh , we're talking about the first category of businesses, some leaders of the category, uh, are starting to build on top of World. We announced this in April of this year, but Tinder uses World ID to prove that there is a person behind that account. Uh, docuine uses this too. Zoom offers a very interesting use case in Zoom meetings, proving that someone is a real person on the other side of the camera.

Octa, Versel, Van Egg, as well as many large companies and enterprises, use it. And that led us to the conclusion that everything was okay. We need to go through phase three and expand this to all the businesses we can get and make this product as reliable as possible as reliable as possible . So, I'll start by talking about a stellar product called World ID. It is mostly used by installing the SDK ID kit for developers, which I will start talking about later.

So, World ID simply allows you to prove that you are a unique person without revealing who you are or anything about you, as we do not have any information about you. By the way, the verification process for a person is quite simple. If you don't know what that is, you can find us on the first floor. Uh, and uh, we can talk about how it works with the sphere and other things, but it essentially gives you unique evidence on a person without revealing any information using CK t technology. And you can use this ID in every application that integrates our SDK.

Uh, and you can have three different credentials if you don't want to go through the realm and do that whole process. Uh, there are not-so-secure credentials called, uh, we call them passport credentials, where you can perform a password check. Of course, everything, it all uses the same principle: CK proves that you are unique to something, uh unique to something, uh , someone, right? Uh, or proves that you are a unique person, proves that you are human, at least to someone, that you are human; it doesn't prove that you're unique because, uh unique because, uh , I mean, I have two passwords, so that makes it unique. Uh, and the face facing the face, uh, is the least reliable or the least trustworthy.

So, essentially, it uses, uh, Face ID from our phones to verify this image, uh, simultaneously with your actions. Of course, it's not as reliable as the Orb, because the Orb is a very, uh because the Orb is a very, uh , interesting device. Uh , interesting device. Uh , this is a super cool camera that takes very detailed photos of your iris. This, uh your iris.

This, uh , Face ID takes a picture of your face. It is not as accurate as Orb. as accurate as Orb. Er, IDK integration is pretty simple. You just need to create an app on our developer portal.

You need to get confirmation of this from World ID and then verify either on-chain or in the cloud. Uh on-chain or in the cloud. Uh , if you don't want to use onchain, that's fine. We have a whole system for this. Uh, I missed this part, but ID kit supports a lot of different programming languages ​​and frameworks, uh, almost all of them, and if you're programming on something that we don't support, please contact me and I'll make sure our team works on it.

Uh, and then yeah, this thing, sorry, I'm going to, uh sorry, I'm going to, uh , OCD was killing me. So, I'm going to talk about this, uh, cool new SDK that we've called Agent Kit. So, agents investigate traffic, as I said traffic, as I said . This is a really, really cool schedule that we've been using, uh, since April of this year. This is traffic that agents have been driving since last year.

He has grown significantly. And, essentially, the way we phrase it, and the question phrase it, and the question we're asking, is a wallet that can't tell whether there's a real person behind this interaction. Well, Agent Kit is made for that. So, Agent has three different things that we already know and have with agents. They have a wallet, agents have a wallet and use it, using XRO2 to interact with them.

Agent Get adds a world ID to that agent wallet and signs that agent into that agent ledger or registry for agents that matches people with agents. Good. So, I as a person can have multiple agents reporting to me or using my world ID. Good. And there are three main capabilities that, uh, these agents can use with Agent Git.

The first is to act on behalf of the person. So act on my behalf. Ask things of a person when the person is in a cycle. Uh, so I use the same usage every time I explain this. But imagine you're building, I don't know, a trading app, and you don't want your agent to trade more than $1,000 without your permission.

You can, uh, have your agent ask you and prove that you, uh, for prove that you, uh, for ...prove that I am acting on behalf of this agent to fulfill this transaction, and you can also, uh, do agent e- commerce or commerce, uh, with the Shopify integration that we have, having your agent transact on different, uh, websites that integrate Shopify, of course, uh, how it works on the website side works on the website side . So, this is how the agent and the human interact, and this is how you, as developers, create things for agents to use. So, essentially , the agent wants...Yeah, that was a little petty.

Sorry about that. Uh, but Sorry about that. Uh, but essentially, the agent client is requesting something on this website using the agent kit. The server responds to this agent and 402 with this call. Um, the agent is signing this call with this wallet you gave him, uh, or, sorry, not with it.

Um, and then the server checks his signature, he looks like the person in the agent's book. Good. And everyone is happy because this agent is supported by a person. Good. Are there any questions about this ?

I-I-I mean, we can exchange messages if you want. Uh, yes. Stop eventually. With all the support. Yes.

Yes, that's cool. I think it's literally...Oh, no literally...Oh, no . This is like...

This is like...This is two slides ahead, but, uh, super quick answer. I mean, each agent is tied to one person. So if you're a person who has like 100 million agents, that website might throttle your speed because they know that those 500, 5,000, or a million agents are supported by you. But let me do this.

So I think it would be better if I just finish this and then we get back to the questions. So we'll continue with this. But I thank you. Uh, sorry about this messaging. I...

I messaging. I...I ...I'll move on to questions right after the presentation. It's almost over.

Hmm almost over. Hmm , so we created three different access modes for Agent Kit. The first is free access to things on websites to things on websites . It's pretty simple. If your agent is supported by a human, that agent never charges for anything , or sometimes you can access a free trial as an agent.

And the third is providing a discount or a certain amount of discounts to agents supported by a person. You can see all the integrations we already have with Agent Kit in the router.world tool. We have integrated with the excs browser search engine, manas research, etc. manas research, etc.

It's interesting to see what's available now. You can also integrate this MTP into your agent and access all of this. So the unit of trust is a person, not a wallet. That was my first answer. That is, I can launch five agents and then ask them to share my trial channel versions and rate limits.

Uh, so rotating more agents doesn't do anything for me because it's all about me. Uh, this is the whole anti- abuse model that we're thinking about with our Agent Kit and the third, uh, SDK that we have. And this is very, very new. I think this is possibly an internal beta internal beta version of ProofKit. So ProKit version of ProofKit.

So ProKit is an open source Rust, uh, rustl, as a DSL, for writing, uh, zero- disclosure schemes, uh, using Noir, uh. And so, essentially, Noir compiles all the math for your circuits and lets you prove things, uh, with very simple, uh, interactions. So basically you, uh, what ProofKit does is it compiles more standard R1C schemas. He proves this with strange evidence, uh, on the client side. So in the phone, uh, it's also, uh, post-quantum secure.

Uh, it's something I don't really like to use that word, but, uh, I think we can, we can use it. Um, so basically how it works is you write your own Noir scheme, or you can use an existing open source one, like, I don't know, CK Password, CK TLS, or CK Mail, whatever. Uh, you can use your own logic. You can prove it on the client side, uh, or the browser side, or wherever, and then you can check anywhere that this proof is secure and, uh proof is secure and, uh , probably true from a probability perspective. Uh, this is a really cool new thing that we just released, and I think this is probably going to be something that World ID will use in the future, uh , to prove a lot of different kinds of authentication.

Uh, so I encourage you to try this, maybe not at this hackathon, but later hackathon, but later , uh, it's a really cool resource for doing client- doing client- side CK proofing . Uh, so just to wrap this up, uh, I just want to give you a couple of case studies that I think you'll find interesting to motivate yourself with. Uh, the first one is the Shopify integration that we have, uh, with the agent set. Uh, we're out of hats right now out of hats right now , but yeah, uh, at the previous hackathon we had a lot of hats, uh, for people who, uh, prove that their agents were being wiretapped by humans, and they can get their agents to buy this unique human hat with a world ID. Uh, that's very easy to do.

You can literally have your agent do all of this for three different teams, uh, by setting up an agent kit that proves it, and then, uh, tell the agent, the outside agent, of course, to buy the product, and it will be completely, uh, outside of the agent, not outside of you. Uh, I guess you can still do it and get a hat rejection. Uh, if you want, if you want to prove it. Uh you want to prove it. Uh , then this is a mini-app , then this is a mini-app .

Of course, it uses a global identifier, uh, and mini-apps as a distribution method. They also have a website where you can check their numbers. It's called credit.cash. So basically, these guys provide, uh, secured anonymous credit to anyone who is a verified person.

So, it's kind of like getting free money because you're human. Uh, and if you pay off your loan, you increase your credit score. And as your credit rating increases, you can get more and more money. I mean money. I mean it starts at 532.

But they get a lot of users. Uh, last time I checked, they had about 1 million users, 1.2-2 million loans issued, and about 2.3 cumulative, uh, cumulative, uh million dollars in loans returned. So, uh, this is a super interesting use case that people have been using and are using, not only how it started in Argentina, but then they expanded to, uh, Malaysia, Thailand , etc.

So using a world ID to prove that you are a unique person. If you repay your loan, you increase your credit score, but if you don't repay your loan, you're out because you can't have a second world ID, right ? Uh, the third one is Morpho. I mean, I mean, everyone here knows Morpho, maybe something like...Yeah.

Okay, great. Good. So, Morpho is a place where you can offer, lend, or borrow money. And we have integration with Morpho, where unique people have increased income. Again, the primitive of a unique person may have a unique advantage.

For Morpho, it's more revenue, and I don't know, uh, getting USDC, and Moro got 465,000 users in the first few months of implementing this interaction, and finally I checked again, they had 91,000 weekly active users using it. Of course, this was only for the first thousand users who made a deposit. But then again, I noticed getting something in exchange for being a unique person, in the age of artificial intelligence and its fury. And the last example is the last example is Tinder. Of course, Tinder uses a global ID in the US and Japan to give unique people a unique match.

And if you do that on Tinder, you get five boosts, I don't know what boosts are, but I think it's some kind of pro feature that allows you to get more matches. So, uh, it's really interesting that Tinder is using this to prove that someone is the person on the other side, and it's the same person that was, uh, confirmed. Um, yeah, another use case for this. Um, the last slide is "go ahead and build." There are four different tracks.

Agent G is someone who has 70, uh, 7.5 thousand, uh, for human use, similar to an agent's. We want new use cases here, not the same things I told you about, like browser-based integrations or something like that. Create things, create cool things, and integrate them with Agent Kit so that unique people can delegate their World ID to agents and do cool things. Um, track B cool things.

Um, track B is the World ID usage options. Track C is for project continuity. So, projects that integrate World ID, but this project has already been created before, and the last track is Proof Kit. Of course, you can use the Mini Kit SDK to distribute your application, because the most important thing for us here is that the projects continue after this hackathon. Um, we like projects that, uh, won't die after the weekend.

Um, and we would like to support these projects after this weekend. So even if you don't win our track, but you think we'd be interested in your project, please stay in touch. I would like to continue communicating with you and help you spread your project. These are all the resources I listed on the listed on the ETH global website. Of course ETH global website.

Of course , you can take a picture of it. Of course, you can also go to the can also go to the ETH global website and get these links. And if you're too lazy to do that, I'll just leave it for a second so everyone can take a picture of it. That's all. If you don't want to do that, you can simply scan the QR code for the documentation.

That's it , I have time for questions. So, yeah, I think you were the first, right? I think you can do it, right? Let me see right? Let me see .

One, two, is it on? One, two, three. Oh, that's it. I remember talking to someone at ETH about privacy issues in the world, and he assured me that this kind of identification only becomes a problem when you can match multiple instances of a particular person's record. So, for example, seeing that this same address was at this point where the bot purchased this item bot purchased this item , and then went to this place at this time, ultimately, it allows these identifiers to be, well, I mean, cross-referenced across all of these identifiers, which ultimately allows us to, let's say, anonymously tie these anonymous identifiers to a particular person, and I've been assured that the world works in such a way that each use case of its identifier is completely independent of the others.

So I was wondering if there was a system where we could see that websites could see that different agents were responding to the same person. How can this avoid the same problem? Imagine I have a website and I get 5,000 agents contacting me. Okay, all of these 5,000 agents come from one person. I can know that.

But the secondary website cannot do this because information is not passed between websites between websites between any clients. Okay, I see what you mean. OK. So it's like an instance on a instance on a website. Yes.

Perfectly. Thank you. So you can check if it's a real person, but can you check who it is? No. In fact, nothing is known.

There is actually no information on our site to verify who this person is. For example, during the verification process, all images taken from the orb are sent to your phone as backup evidence and then deleted from the orb. So, if you don't trust me, right ? You may know that the entire repository is open source. So if you can, you can go to the worldcoins repository and check out the orb firmware.

You can check out the iris repos repository. So everything is open source. If you don't trust me, of course you can check it out. This will certainly encourage you to do so. Uh, but yeah, we don't have any information about the user right now.

But how do you know I'm not using myself to verify about 10 different accounts? Because the biometric data of your iris is unique to you. So if you try to verify yourself in Orb again, Orb will say, "Hey, you already exist in my database." But you said you don't store images, we don't store images, but we have a hash that we call the Irish hash. Essentially, it's a hash that proves you've done the verification.

So if you try to check again, this hash will probably match because you have the same iris, it will probably match the previous one, and you'll say, "Hey, you already exist here," and if you change your iris , it's almost impossible. I think I know that in the future, contact lenses will also be recognized by Orb, but like colored ones, of course, yes, of course, yes, they, I mean not by Orb, but by AI models that work to get this, uh, CK, uh get this, uh, CK, uh , this, uh, hash from the iris. Yes, cool. Thank you. Yes.

Uh, yeah, we have about five more minutes, I think. Uh, I mean, in the noir schemes, I saw that it said " said " post-quantum". Can you elaborate on this a little more? I know it's taboo. The word " taboo.

The word " quantum" is taboo in this field, but I think it will be a long conversation. We could go and chat downstairs with the team because that would be fun. Oh, we're out of time. Oh, okay. We can chat downstairs at the booth.

Uh, this will be fun. Yes. And I think we're out of time, but I can stay here or go outside and answer answer any questions. Yes, I will be at the booth all weekend.