Hello everyone, my name is Mateo. I am part of the TFH team. We create tools for Worldcoin tools for Worldcoin . You may know this project by that name, but we have renamed it World. Just World.
And so, the idea of this seminar is the idea of this seminar is to explain our new products to you. We have several new products that can be implemented using World ID or ID kit. This is one of our three SDKs. And the third direction that we have direction that we have is agent kit. This is another of our SDKs that you can use to do interesting things.
But I'll get into that later. I always start my presentations with this photo. You may know this guy, Peter Steinberger, who created Open Claw or ClawBot created Open Claw or ClawBot , or whatever you want to call this product. And that was the moment my brain thought, "Oh, wait." Are you Frederick?
Oh, no. Dude, I thought I knew you. This guy literally appeared on one of Lex Friedman's biggest podcasts and said, "I'm watching my agent happily press the 'I'm not a robot' button." And I thought: " robot' button." And I thought: " Okay."
I think people will now understand that World ID is useful for the future of humanity. It happened this year . And a lot has changed for us as a company over the course of this year because of agents in general. So, there is a need for proof of humanity. In fact, there are some outdated statistics— statistics— the number is probably already higher than 50% the number is probably already higher than 50% —but over 50% of —but over 50% of internet traffic now is robots or agents doing things for us.
CAPTCHA, as Peter said, doesn't work as well as it did before agents came along. Actually, it does n't work anymore. KYC and identity verification are not working in the best way. You may have seen many deepfakes on various platforms. So there is a real need to know that there is a person on the other side of the computer.
And this is the problem we are solving. Okay, we know you're human, and we're proving it to the internet without revealing any information about you, just confirming just confirming that you're human. Of course? This is the core and basic idea of the entire protocol. So yes.
World ID is the name we gave this thing. This is a high- strength authentication level for protection against bots. Eh, it provides privacy. You can build on it . You can use SDKs, which I will mention later.
And yes, it uses ZK- uses ZK- proofs for everything you attest. So, no information is disclosed. This is open source. So, if you don't trust me, you can check it out yourself. And if you don't understand what's going on, you can read the repository and see what's written there.
Or ask your agent to do it, whichever you prefer. So, yes. You can rely on everything that is public on everything that is public , or listen to this talk and make your life easier. So, yes. World ID has three different credentials that we want you to test in this hackathon.
I use levels to make it easier for people to understand. The lowest level of assurance is face verification. A feature that we recently released, which is still in which is still in beta, and that you 'll be using this weekend, is called "selfie is called "selfie check." It matches the image stored on your phone with the real-life verification during the selfie process during the selfie process . This is the method with the lowest guarantee, but with the least effort least effort .
Okay, the second one is a passport. You know that your passport has this NFC your passport has this NFC chip. You can attach it and it will provide a lot of details if the government allows it. Some of them allow it. We work with many governments to allow the use of passport data.
So you can also use your passport details to prove that you are a human. At least that's what we trust we trust —governments say you're a citizen of a certain country. That's great, but it doesn't mean you're unique. Because, for example, I am Argentine, I have two passports. I can do this twice.
I don't really like it. But we have the highest guarantee method—it's an iris scan or a level of verification through Orb. This is actually a visit to the Orb. We have two here. You need to take off your glasses, scan your eye, and prove that you are a unique person.
Because if you didn't know, your iris, like your DNA, is the only biometric evidence of your uniqueness. And they uniqueness. And they ... Oh, sorry. And they cannot be faked.
And not your DNA, not yet. Yes. We used to say "never," but now we say " "never," but now we say " not yet." And we will probably solve these problems when they arise, because we are human and that is our nature. But yes.
Um, sorry. Um, I'll get back to the topic and the presentation. For EthGlobal in Lisbon, we have three different directions. New uses for Agent Kit. I'll dig into this, break it down in detail.
8 thousand in prize money, divided between the three best teams. The second direction is beta direction is beta testing of selfie testing of selfie verification. 3.5 thousand for the two main winners. And the third is beta winners.
And the third is beta testing of personality assessments. For selfie verification and certifications, we allow existing projects to build on them, because these are the areas where we need your feedback the most. In fact, feedback is part of the results for both of these areas. Selfie these areas. Selfie verification and personality attestations are currently in beta, but we will open them up to all of you this weekend.
I mean, for everyone, but you're hacking here. And if you still want to use these products, contact us and we will add you to our premium beta testing developer group. Good? But yeah, let's move on to the actual part of the workshop. Agent kit.
Agent kit the workshop. Agent kit. Agent kit is essentially a way for you, as a human, to delegate your humanity to your agent, okay? So, you have an agent and you can provide him with a wallet, okay? And when integrating Agent kit, you can delegate your humanity to this agent so that it can perform tasks for you.
For example, he can go to the Shopify marketplace and buy things for you buy things for you . And if this Shopify marketplace is integrated with us, the human-backed agent can get a preferential discount. Because Shopify wants to work with agents who are backed by people. This is one example. There are many of them.
And I look forward to you creating the rest that we don't know about yet. It's a very seamless integration. And two more interesting use cases that I think are worth understanding are network delegation. "Human in the loop" means that you, as a human, delegate your World ID to this agent, but you don't really trust him, you know? Therefore, you implement human control.
So human control. So , as the name suggests, a person intervenes in the process while the agent is doing something. A simple case or example: I entrusted my agent with a wallet, gave him money, allowed him to trade on Hyperliquid, but I don't want him to spend more than $1,000 without my knowledge. So in this case, I add a person to the decision-making process. And again, there are many other uses for Agent Gate.
This track requires meaningful use of Agent Gate: verifying that an agent has human support before granting access, working on an end-to-end process, and new verticals, in addition to these obvious examples I just gave, and many others. A few more examples of new verticals for human-powered agents: pricing, fulfillment, accountability, etc. I would take a picture of this slide if I were creating a set of agents. Agent workflows that combine workflows that combine , for example, identification and payments. These are ideas that were in my head and I just wrote them down.
Of course? New trust flows between agents, agent-API or agent-something that you need to work on, or new regulated workflows for human-supported agents, again using a human in the process. This is an integration with Shopify. She is already working. The caps have already been distributed.
Some have already received them. But it's as simple as asking Claude, "Hey, help me buy this product on this website." And since your agent already has delegated authority, he can do it on his own. It's pretty simple for the user, and just as easy for you as a developer to integrate it. What wo n't work.
Oh, a lot of things. Because we've seen many different use cases. I won't list each one individually, but in general, simple reputation systems for agents, we've already seen a lot of them. Agent interaction with human support and simple content generation—we've seen plenty of that already. Benefits for agents with human support, like, "Oh, if you're an agent, you get five free API calls because there's a human behind you."
We have already seen many such examples. And all the others that I won't mention because time is limited. But overall, I think...if you have any questions about what will work and what won't, just reach out to us. We will be at the stand, agreed?
SelfieCheck. Actually, SelfieCheck is the easiest way to join World ID. It simply uses, as I said, a mobile camera to establish the humanity or "aliveness" of the user. This gives a civility score because it's a low-level assurance method, we want to tell you, "Hey, this user or this set of numbers has tried to access your app many times." So, in a sense, it's a way for you to detect if someone is trying to make a lot of login attempts into your app.
So, the civility score is something interesting. This is extremely fast. I mean, everyone here, maybe not everyone, but most of you have been tested for "live presence" in fintech apps and things like that. It's literally the same user experience for the user. And for you, it's just ID Kit integration or your agent integrates ID Kit, whatever you want.
And yes, this is the lowest level of guarantee, but at the same time the least burdensome option we have right now. Regarding track requirements, you must integrate selfie verification. You should cover both depth testing and user feedback for this track. You need to have a working prototype or something that works. And so.
If you want, you can get this clear report by separating user feedback and depth testing. It's part of the same thing . Regarding the desired feedback, it is not mandatory, but if you want, we would really like to receive it. I won't mention it, but just tell us how the integration went, whether everything went well, what is the value of this for your product. This will help us understand and test your product.
Examples for this, I mean, a lot of the things I mentioned are not things we don't want for Agent Kit. Since it's brand new, it all fits here. Free trials, anti-spam trials, anti-spam onboarding with this score onboarding with this score we have. Continuity checks, "live presence" checks. You can use Selfie Check to check for " Check to check for " live presence", for example, whether it's a real person.
Again, risk-based flows. So, if a user is verified only through Selfie Check, we can give them one thing can give them one thing , and if they are verified through Orb, we can give them another. etc. And finally, identity verification. So identity verification.
So , as I said, we have this passport thing. And this passport has many attributes. The only attributes we can use right now are age, nationality, document type, country of issue, document number, and full name. So, essentially, as a developer, you can ask on your platform, "Okay, I just want to, I'm Argentinian, I already told you." I only want non-Argentines because everyone hates Argentines right now.
I don't know why. We lost, man. I don't know. I do n't understand it, but I don't care. We don't want to see Argentines on this site.
So you can use identity verification to keep Argentines out of your site, for example. Or you can use identity and selfie verification so that there are no Argentines, but only verified selfie verified selfie users. You can combine both methods and apply in both directions. This is the point I wanted to make. Yes, it's actually a very easy integration.
The only answer from us is that the criteria were met. So if you only want users over 18 , our answer would be simply: "over 18 would be simply: "over 18 ", and that's it. You won't know the user's age, okay? The direction requirements are the same as for selfie verification. So if you want, you can take a picture of this, but it's literally the same as the previous one.
I'll wait until he comes back. Perfectly. And so, the evaluation criteria will be divided as follows . Essentially, it's the quality of the product and its relevance to us. 25% are reviews about face verification and selfies.
So, as I said, don't be too merciful. We really need to test these things. You need to tell us what's bad and what's good so that this product can appear in the future. 20% is the quality of the product itself. 15% the product itself.
15% —technical integration. So if you use an ID kit, you need to verify the evidence on- chain or in the cloud, or wherever. In a cloud, like "clouds". Good? You need to verify them.
We will check this during your presentation. Of course? So make sure, or ask your agent to make sure, that they verify the evidence. I don't know why I call him "he," but I don't care. He verifies the proof.
And yes, 10% is whether the product can be launched and continue to work after the hackathon. In conclusion, I told you about Agent Kit. We want new workflows and features for Agent Kit. Selfie verification— Selfie verification— the lowest level of assurance, minimal effort. Feedback is what needs to be provided.
Personal certifications. Again, you can ask questions and restrict access to Argentines on your websites. Feedback can also be obtained. And lastly: winners will receive priority access. At the very least, I'll be reaching out to them and " them and " getting" them for 30 days, trying to help get their product out into the world, and maybe if they want, I'll invite them to our accelerator or one of our development programs that we have right now .
That's all I have . People online are a real foundation . And I want you, if you want, to scan this QR scan this QR code and go to the documentation. We have MCP for both documentation and the developer portal. So you can ask your agent to do all these integrations for you.
But yeah, basically the only thing you need to do is register on our developer portal. This will give you a team, application ID, project ID, and with that you can start working on your projects. Thank you.
Automatic transcript — names and jargon may be misspelled.