Bootstrapping Global Scale Proof of Human PoH with Self I Rene Reinsberg

We are ready for our last session of the day. And for this one, I'd like to bring on Rene from Celo to talk to you about how they're bootstrapping the entire graph of verified proof of humans through the self network. So, without further ado, please give Rene a big round of applause. Thanks, Kartik. Wow, this is I mean, probably the most beautiful Pragma venue yet.

You guys really stepped it up. Love it. Awesome to be here. Let's see if the clicker works. Yeah, let's talk about bootstrapping global scale proof of human.

A show of hands, proof of human, is that a term that you guys have heard of? Yeah, okay, I see some hands. You're not alone. Mark Andreessen tweeted this week, and I think it got a lot of people thinking more about what does it mean, why do we need it, why do we need it now? And I wanted to share some insights from our journey through my time kind of starting Celo back in the day and everything we've learned about identity doing that to to now running Self, which is a spinout from from Celo that we started about a year ago.

Actually, a bit more than that. And it's pretty interesting because I think a lot of the lessons in some ways similar to the talk just now that we sort of, in a way, hardened in web 3 now become suddenly applicable to many more use cases outside of web 3, you know, in this new sort of agentic web. And I think that's amazing. I think in many ways, I forget who tweeted it this week. There was I think it was a Seaf, maybe I'm giving credit wrongly here, but um you know, I live in San Francisco and crypto people there are generally like doing their thing, shugging along, but the AI people now are becoming really bullish on crypto because they're discovering some of the stuff uh we've built and how it can be used in the context of what they're trying to uh do and get really excited about stuff that we've been excited about a year or 2 years ago.

Um and so I think if that's a predictor of where things are going, I think it's it's really good news because a lot of the stuff we've built is going to become much more widely applicable. Um and that's part of my talk here as well. Um All right, so there was another uh tweet or another announcement this week. Um this one is a little bit Oh, here we go. Uh little bit closer to to home.

So, this is a uh something we're doing in partnership with Google and Tether. Um and uh yeah, in a nutshell TLDR, uh Google is using Self to give people free money. And it's actually a really big deal. If someone had told me, you know, even like a year a few years ago that at some point soon, there's going to be a new product on Google where you go and you verify that you're a unique human and you can basically claim free money, send it to your wallet, and start using it, uh I would have been like, "No no way, it's not going to happen in this decade, maybe not even in the next decade, maybe ever." Um but it's it's happening.

We announced it this week and it's actually uh going live next month. Um which is yeah, it's just nuts. Uh this is kind of an early mock, so it may look slightly different than that. But the basic idea um is to say, "Hey, look, you know, what is some of the friction people face when they're when they're onboarding, when they're kind of trying to use stablecoins, right? There's a lot of friction on actually getting your hands on them.

And we wanted to just make that super easy. Now, the nice thing is Google um a few months ago already started using self for their testnet faucets. Um so, you may wonder, "Okay, why would they be interested in that?" Turns out, there's actually um yeah, quite a bit of a secondary market around some of these testnet tokens. Um and so, in order to prevent people from just kind of abusing their system, they kind of started introducing that kind of proof of human there um to make sure that similar to, you know, like if you're claiming an airdrop or, you know, kind of going to get a token distribution somewhere, um you you kind of want to make sure that, you know, you're not just handing out tokens to a thousand people or like, sorry, a thousand bots or wallets that all belong to the same person.

The idea here is to have this kind of proof of human, which allows for ensuring that, you know, you know, only one user uh can actually at a time uh claim whatever you're giving out. So, this is pretty insane. I think it's actually like going back to very early Celo days, 2017, 2018, what got what motivated us to to start building Celo. We were like, you know, there is the potential for a better financial system, for doing things, you know, regenerative sort of financial mechanisms like UBI and doing them at global scale. And the fact that in 2026, I'm standing here and talking about this is kind of nuts.

Um so, yeah, pretty stoked to see this roll out and people everywhere in the world being able to uh yeah, get some get some free money. And you could imagine where something like this kind of goes longer term, right? Where um Um, you know, you're thinking about even like social services for entire countries being, um, you know, kind of rolled out more efficiently using crypto rails, but then also using, um, you know, cryptography for identity to kind of prevent fraud. And, um, yeah, just streamline this operationally, which, you know, it's it's a pretty big deal. For anyone interested in learning more about this, there's also a case studies.

So, if you scan this QR code, Google wrote a whole case study on how they're using Self and how the tech behind it, uh, works. Uh, so, that's reading for your plane ride, uh, back later this, uh, or like after the weekend. Um, and yeah, so, wanted to use this opportunity to actually dive in a little bit and talk about, you know, why is this becoming such a big thing? Why is proof of human suddenly something Mark Andreessen tweets about and, you know, why are we building Self? And, you know, I think this has been true for a long time now that, you know, we've we've probably been wanting a better solution and, you know, I think there's all these use cases, some of which I mentioned in crypto token distributions.

Um, but but I think fundamentally, I think there's there's maybe even, uh, a bigger need when it comes to the internet itself, right? When you think about where you spend your time online and, you know, whether, you know, even like Reddit was mentioned. Um, yeah, how do you know right? Um, and today this is more, uh, more urgent than ever. Like, is this something this is a bot account?

Is this a real human? Who is this human behind this content? And there, you know, have been all kinds of different approaches to to solving this this problem at sort of internet scale. And, you know, some of them have worked in in sort of a small scale. Some of them maybe are are still sort of actually being explored, but um the truth is today we we don't have uh a real solution yet that can reliably actually give um sort of proof of human infrastructure at at global scale.

And the the thing that um was for us sort of an aha moment um actually, you know, about 2 years ago starting to chat with some folks who were doing research uh on this. Remy, one of them is in the room. Um and is uh since, you know, kind of joined forces uh with with Self and kind of joined the team as a co-founder. Um you know, had this insight that we already have, you know, this kind of really robust hardened infrastructure out there. It's put out by governments.

And um you know, if many of you here may have like maybe European ID and you see this little kind of sign like this bi- basically biometric chip um or a passport that has it. Um and you realize, wow, there's, you know, already billions of people around the world that have um you know, a um you know, pretty incredible uh mechanism for um for proving their identity. And you know, so far like well, up today, this is not really taken advantage of other than when you travel and you're, you know, kind of at an airport and you're sort of entering a new country and they're um they're taking your passport, right? Or your ID. Um but it's actually an infrastructure that's public, that's open, that can be used for many other things including um identity on the internet.

And so, without going into like all the details, but you know, wanted to share this kind of high-level um the way it works is for each country, you have this country signing certificate authority that signs these document signing certificates, which in turn then sign passports. Uh Um it's about 100,000 passports at a time, and then the next certificate kind of um becomes active, and so on. And so, if you actually go and kind of look at this infrastructure for each country, um you can enter building and developing a system that can prove in zero knowledge that a certain ID or passport is indeed part of that kind of collective of valid passports. And that's really one piece of the power behind Self uh doing that and allowing people to first of all, kind of when they register to prove that they're um you know, that they have sort of a valid document. And that's done through NFC scanning on your phone, so on your mobile device.

Um so, from a user experience, actually pretty seamless. And also, you know, you may be kind of when you hear this, you may be thinking about, "Oh yeah, you know, I like KYC'd for like this service the other day, and I had to hold up my passport and turn my head. Is it similar to that?" Well, sort of, right? It's that's kind of, you know, KYC.

It's a process that is most widely used today to verify someone's identity, but some of you may kind of wonder already, "Well, is this going to work in the future, where it's becoming incredibly easy to, you know, create sort of fake, not just photos, but videos?" And you know, really what you what you're lacking there is to have sort of this kind of true, you know, kind of in a way like uh machine-read proof, right? You're not relying on a picture, but you you're actually through the NFC scan um proving that the authenticity of the document is is is real, right? And that's that's pretty incredible. Now, the second step is that um or actually, yeah, sorry, expanding on that a little bit, um is you know, the way kind of Self uh works is it's actually kind of, you know, using zero knowledge technology to do these proofs so that you don't really so you don't have to trust Self, right, for that attestation for us like, you know, kind of, you know, reading the passport and storing the data somewhere.

Um, but not requiring a third party. And the benefit of that is too that we're not actually storing any of the user's data. Um, and that's, you know, I think in today's world also a big deal where every every day, every week, you know, you read about some of these um, big hacks of KYC companies. So, once the user is registered, then there's the second part which is the really fun part now. Any application can actually request proofs about the data that the user has now registered with.

Um, and and request it in the form of these zero knowledge proofs, right? So, rather than um, the application saying, "Hey, send a picture of your passport or ID." It can say, "Please send a proof that you're from Germany or that you're over 18." And the, you know, Self app then kind of creates the proof and shares that proof with the application. Which is nice because the application now also doesn't have to store any PII.

And, um, can rely on the proof um, to go about its business. Which for many, you know, kind of applications, that's, you know, highly preferable over actually kind of requiring the user to go through a KYC. And of course for the user, um, it's less cumbersome as well. And, um, the nice thing is once you're registered doing this, it takes a few seconds, right? So, it's, um, it's really it's a really elegant experience.

Now, you can construct all kinds of, you know, functional proofs here. I think one that's pretty popular whenever money is involved is checking whether someone is on the OFAC list or sanction lists. Um so, that's that's really powerful. But, also we've seen more kind of fun things like, you know, back when meme coins were a thing, people uh you know, launching meme coins that only people with a certain first name uh were able to claim. Everybody else had to uh buy into them.

So, yeah, some enabling a bunch of a fun new use cases. Um so, those were the two steps to repeat. So, first one, registration. Second one, you know, allowing any application to request these disclosure proofs and then the user being able to share those either from the self app or we also have an SDK which a bunch of bigger um web properties are integrating now natively. So, you know, you can also then soon use it directly in those.

Now, a big question is, yeah, this is great. I get the web3 use cases, but, you know, how does this uh play into sort of the bigger the bigger web and like why at this time was was there such noise about it? And so, I want to give a little bit of a foray into into sort of the agentic world. Um and uh yeah, I think one interesting example, the MoatBook, you know, which since actually has been acquired um by by Meta. Um and, you know, it's sort of the the premise here, it's a uh social network for for agents.

And you know, you kind of when you enter, you kind of, you know, prove that you're where you you you know, you're an agent. Um and then you look at some of the numbers uh there in terms of like, you know, followers and kind of engagement. And you start to wonder, well, okay, um how does that translate to to humans? You know, is it maybe uh you know, a handful of humans that have just spun up like a few million agents that start talking to each other. Is it, you know, one or two agents per human maybe?

You know, what's sort of the what's the backstory here? And and so, it's suddenly like um proof of human or sort of thinking about proof of human in the context of agent becomes pretty interesting. Um even from a uh yeah, kind of almost like a kind of agent civil perspective. Um and I wonder remember what I kind of pointed out there. Yeah, just kind of seeing the engagement on some of these uh posts uh yeah, the numbers kind of like yeah, didn't really make sense.

So, you're kind of really starting to wonder um yeah, what's what's the actual engagement here? What's the actual number of users? This one is really interesting, too. Uh Rent-A-Human, who here has seen it? Anybody on it?

Anybody used it? No? Um so, the premise is, okay, there's going to be a bunch of work that agents can't do themselves, so they're going to hire humans to do it. Now, of course, uh be kind of silly if, you know, like an agent is trying to hire human, but maybe the human is not actually a human, it's just another agent pretending to be human to get the reward. Um so, those agents really need to know that, you know, the counterparty is actually human.

Uh and that's the premise behind this this kind of marketplace. And yeah, there um I think when they started out, they were charging people $10 a month uh to basically verify that they're human. Uh and again, you know, it's interesting cuz like that's in a way putting a market price on being human in sort of this agentic age. Um but it it points again to a to a bigger challenge, right? Which is, okay, a lot of workflows are getting changed.

There's like questions around how How we delegate our identity, you know, to agents or maybe, you know, we all I think this is a lesson many of us have maybe already painfully learned. Uh LLM's agents, you know, AI kind of, you know, is bad at keeping secrets, right? So, yeah, maybe, you know, um yeah, some is a some tears in the room here. Um but uh yeah, maybe I don't want to give my agents like all my personal information, right? So, it can do more useful things.

Um and I think that again this is something where self we we think self can really play a big role in, for example, as an agent kind of navigates more complex workflows, hits sort of a barrier right where it requires to, for example, prove that the human they're representing is from a certain country or, you know, is, you know, not on the OFAC list or maybe is over 18, um then it's really simple to trigger a request to the human, right, to kind of share that proof with the application or the site that the agent is or working on, right, in that workflow. Um and then kind of unblock sort of the agent from kind of being able to continue to do the work without the human having to actually, you know, share the actual data with the agent. Um so, pretty interesting to think about some of the product solutions this can enable that some of you maybe building in the context of, you know, agentic commerce workflows, you know, just general delegation um in this kind of agentic world. And I think that's like it's kind of crazy. It feels a little bit like, uh you know, early early crypto days where it's like, okay, we're we're we're starting with just like building these simple things like, how do I move, you know, like money or like a token from person A to person B and replicate sort of like a payment, you know, like before to then kind of DeFi happening, other things happening, right?

Like there's I expect similar things here where a lot of the early use cases are still mirroring things we, you know, have already been doing, but maybe we can do them faster, more efficiently, and cheaper, but, you know, then you also kind of start moving into sort of the native agentic territory, and it's pretty clear that identity is going to be a key bottleneck or and or enabler depending on how you look at it. Um for that purpose, we actually spun up a ai.self.xyz dedicated site for agents. So, further down there are some skills on there, and like, you know, agent specific instructions.

So, if you have an agent, please send your agent uh there. If you're interested to learn more about um self and you're human in the room, hello. Um check out self.xyz for more information. And uh with that, looks like I've hit my my time here, but uh it's been amazing to be back on Pragma stage, and thanks everyone for uh listening, and happy to answer any questions after.

Thank you.