DC Builder / World Foundation

DC Builder on World Foundation's proof of humanity for AI commerce

InterviewJune 2, 202625:08

In this episode

The internet is filling up with bots, and nobody has built a reliable way to tell the difference between a human and an AI agent at scale. PROOF OF HUMANITY is what World is building — and its full-stack upgrade just became the identity layer for AI agent commerce. DC Builder, Research Engineer at the World Foundation, joins Ashton Addison on Blockchain Interviews to break down what changed in World ID's full-stack proof of human upgrade, how the AgentKit launch with Coinbase puts verified human identity at the core of AI agent transactions, and whether proof-of-humanity is possible without putting an iris scan at the center of it.

You'll learn:

  • What the full-stack proof of human upgrade actually changed in World ID — and why it matters specifically because of AI agents, not just bots
  • How AgentKit uses zero-knowledge proofs to link AI agents to a verified human without exposing any personal data
  • What the Orb controversy gets wrong — and whether there is a credible path to proof-of-humanity that does not require a biometric scan
Key takeaways
  • World ID uses custom hardware orbs with cryptographically signed camera feeds to verify humans at scale while preserving privacy through zero-knowledge proofs.
  • Standard smartphone biometrics like Face ID would create collisions at 15-30 million users, requiring custom imaging hardware for billions-scale uniqueness verification.
  • AgentKit integration with Coinbase links AI agents to verified human identity without exposing personal data, enabling secure agentic commerce.
  • The internet lacks a native identity layer similar to how it lacked native payments before cryptocurrency, making proof-of-humanity infrastructure critical.
  • World ID operates orbs across approximately 20 countries including Portugal, Argentina, Thailand, UK, and Germany to increase accessibility.

Chapters

Transcript

Read the full transcript 4,641 words, auto-generated and lightly edited

I'm Ashton Addison from the Crypto Coin Show, and today on Blockchain Interviews we have DC Builder, research engineer at the World Foundation, the team building the world's largest proof of humanity network. They just launched the agent kit with Coinbase to put human identity at the core of AI agent commerce. DC, welcome to the show and thanks for taking the time.



Thank you for having me. Pleasure to be here.

Yeah, excited to dive into this. I think a lot of people have heard of World and they know about digital identities, but they don't know the details or how it's going to fit into their everyday lives and then tie in AI agents into that as well and we're still very early in figuring out exactly how humans and

AI agents are going to interact and how agents are going to interact with each other. You've been working on this with Tools for Humanity since 2022. I feel like you have a really great grasp of this. I would love for you to paint a picture of what is the World ID and Tools for Humanity and what you've been working on for the past few years and how that ties

into what we're seeing with AI agent growth right now.

Mhm. Yeah, it's the world has changed a lot since then. This is pre-ChatGPT. I joined in July. ChatGPT launched I think in November of 2022. So it's been a wild ride since then. So actually funny enough, I was I was working or I was interested in AI as a high school student. It was like 17, 18.

So this was like 2019 roughly. And so I was really interested even in AI before joining the crypto space. And generally like the reason why World was was started or co-founded by by Sam Altman, Alex Blania and everybody else on the team was because they could see forward into the future and extrapolate the pace of progress into sort of something that we see now or

like they tried to predict sort of like what the world's going to look like. And essentially it converges into a world where you cannot teleport AI-generated content from human-generated content. Be that text or videos, images. Maybe if you've seen like Instagram Reels these days or TikToks, a lot of the content these days is generated by AI. And what

Worlds was set out to build was also a way to prove that you're interacting with real humans on the internet and on blockchains, right? Like you want to know that this is this is a real human that I'm interacting with. Even now on Zoom, right? Like I want to know that I'm talking to the real Ashton and not on an AI generated deepfake of Ashton on the

Crypto Coin Show. And so this is a mission that I truly believed in when I was joining. And it also embodies a lot of the crypto sort of technologies and philosophies, like privacy, decentralization, permissionlessness, scalability. And it was a product in crypto that tried to scale to billions of humans, and it's still to this day trying to

scale to like be a true product, which back in the day there were very few products that tried to reach mainstream. So yeah, that's that's a good general introduction.

Definitely, that's a great intro. And it's one thing to be talking to somebody anonymously on X that just goes by some some username or have a blockchain wallet they don't know who

you're dealing with. But then now integrate AI into that, and people are able to fake identity, video, images, everything. And now now I feel like that issue that was the core problem to solve many years back has become an even bigger issue potentially.

Yeah, and it's going to become an ever more prevalent issue into the future. Right now you can still tell if the

video is AI generated. You can have some cues, like look for text that is like weirdly structured. But give it one or two more years, and it's truly going to look indistinguishable. So and the internet doesn't really have a native identity layer, either. As it didn't have a native payments layer either. I like we're we're just now building like crypto

tools that are ready for mass adoption with stablecoin payments for the internet. And now agents are able to use the internet on behalf of humans and they're using the internet at a much faster rate or like they're they're able to do a lot more throughput and interact with a lot more applications concurrently. So things like agentic commerce for payments or like an agent

booking your travel or paying for specific things you want to do or orders, you know, like groceries or what not. And so now a lot of the tools that we've been built in building in crypto are starting to become really useful in the context of the internet and the AI world that is now emerging. And so yeah, the there's going to be a way or there needs to be a way that is

native to the internet for proving you're a real human and crypto and World ID I think play a core core role in this.

Mhm. So with World ID, the focus is having that proof of humanity for humans. Does that just involve uploading your ID or doing like a live video? You know, how intricate do we need to get?

Mhm. Yeah, so to actually so solve

that question in the very beginning we tried to look at all the existing solutions that existed for ID like whether it's passports or existing kind of biometric based the solutions. And there's no solution that was scalable, private, and at the same time decentralized. So we decided to build our own solution which is the orb which you can sorry, on the other

side. So you can see this device right here. So th- this device is what we call the orb and it's able to prove you're a real and unique human in a fully privacy preserving way. No no information leaves the device. And after the user verifies in person, they a verified World ID. And using zero knowledge cryptography they're they're they are then able to create a

proof of human or proof of humanity, which then anyone else can verify, right? So, if you're an application, you can verify this proof, and you learn nothing about the user but the fact that they're a real and unique human.

Very interesting, and I actually received some emails from the team saying, you know, in my city there's some world orbs you can go down and

verify. Can you explain the necessity of physical hardware and going somewhere to do that versus using your phone like people do for KYC on exchanges?

Yeah, so KYC on exchanges usually works in the in the way where they you take a picture of like someone's ID and you store it. So, this violates privacy, right? Like you now have to keep a database of

credentials. This is not what we want to do. In the context of uniqueness, you want to measure that someone is real and unique at scale. And so, the reason for custom hardware is because the orb is a trusted camera. So, it's not something you can spoof easily. It's it's it's meant to be built in a way where you cannot attack it on the sensory side.

So, like the camera feed itself is cryptographically signed. And once you verify locally like all the machine learning models that run on the device to check they're a real and unique human being

[clears throat]

is done fully locally, so no information leaves the orb. And a lot of the for example like the resolution you need to measure uniqueness at scales of billions of

people needs custom camera imaging tooling that is not present in local phones, right? Like if you were to use Face ID in the same way for example from like a regular iPhone camera, you would get collisions at like 15 to 30 million users, right? Like if you have a 30 user-base using Face ID, a new user would try to verify, they would collide against somebody else's ID

because of the way that these systems work. And so, you need some custom imaging, and this is sort of like the reason why it was a very painful conclusion for us to have to build your own hardware because a lot of things go into production, assembly, distribution, and you cannot use the network effects of existing hardware like phones and other devices.



Hm. And with having these physical hardwares, I know they are in the US, what's the distribution like? And, you know, there must be there cost it to set that up, and, you know, have to rent brick-and-mortar places. Like, I thought that people are trying to get away from having to pay for brick-and-mortar places, or how do people actually access these physical orbs?



Yeah, totally. So, we have a website where you can find like all the available orbs. If you go to world.org/find-orb, you can find all the orbs in all the markets. I think it's roughly in like 20 countries at the moment. Off the top of my head, I can remember Portugal, Argentina, Thailand, and UK, Germany. There's a few others. But

the general plan here is to make orbs available in as many places as possible. We've been operating orbs, I think, since 2021. So, we already have like 18 million users that have verified with the orb, I think, roughly that last time I checked. And the goal is also to create other proofs of human. So, we're also expanding into things like

like personhood like like KYC credentials, but in a way where you can use privacy technology to prove things about those credentials. So, you can use zero-knowledge proofs to create a proof that you're over 18 without revealing who you are, right? For example. So, this is a new new version of like how to prove personhood, but it's not as robust as

the orb itself because again, like you can have multiple passports, and it's not a system that was designed to solve the issue of proof of human.

Mhm. Definitely. Well, that's pretty cool that's already available, and there's there's tens of millions of people. What are the use cases for this outside of what people think from the Crypto Coin Show

and crypto exchanges and finance-related applications? You know, I think I saw a partnership with like Tinder and Zoom and things like that. How how far does this expand out to people that aren't involved in Web3?

Yeah, totally. So, I think the main areas or like the more mainstream areas where there's lots of users would be verticals like gaming,

dating, and social media. So all of these three verticals have more than a billion users each, right? Like Meta has like 4 billion users itself like for Facebook and Instagram and other services. For Tinder, right? Like it's also I think like hundreds of millions of users. For gaming, right? Like who doesn't play like Fortnite, Minecraft, or like thousands of other games? So,

as you have these multiplayer games, you have things like ranked gaming, etc. You care about having a proof of human there because you don't want to play against bots. Like you can imagine a world where if you go on Tinder, you can be talking to an AI-generated image of a guy or girl, and they can be pretending to be someone who they're not, and it sort of breaks the thing

that you're looking for when trying to go on Tinder or play a game or talk to someone on Zoom. And so, our recent partnerships with Tinder and Zoom are to enable the real human experiences that you're you're expected to have. In the past, this wasn't a concern because there wasn't really a way to generate things that could fool the human eye, but as AI gets better and

better, now you can no longer trust that this is actually Ashton I'm talking to, right? Like if we if we didn't know each other through other means, then now you have a way, for example, to have a identity verified badge, like World ID verified badge, that would show up somewhere here in the Zoom call to show that both of us are World ID verified and that we look what the

Zoom feed is looking like, right? It would match the image of the orb that the orb once took, that is stored on the user's device, against the Zoom feed. And you can create a zero-knowledge proof on your own device without revealing anything to anybody else. And then this proof would be then verified by Zoom, so Zoom can then show that this is a World ID verified human and that

it's the same human that went to the orb that got verified. And the same for Tinder, right? You can verify that you're you're human verified on Tinder.

Definitely. That's pretty cool. And I want to jump into the AI agent part now, cuz I feel like that confuses things, as one person can have thousands of AI agents and may I don't know if they

can have identities or, you know, they definitely have different tasks, but how does that complexity get addressed with if AI agents start being able to do tasks and almost act on the human's behalf, but in different ways?

Yeah, so one example that I like to give out is imagine you're an artist, like a music artist, and you're you're trying to sell

sell tickets on Ticketmaster. A thing that happens is something known as scalping, where a bot this doesn't even have to be AI, but a bot can interact with a website and buy thousands of tickets, essentially like all the tickets they can have they can get their hands on and then resell them at a markup price on a secondary market, right? So the

actual fans, they don't have the ability to buy a ticket because they're sold out, because there's someone who made a bot that bought all the tickets for them. And so now they have to pay a lot more for a ticket from these resell marketplaces instead of actually just interacting directly with the artist. And so a lot of these experiences are

are going to get worse and worse as AIs get better because all the tools that were used to prevent, you know, like captchas, you know, like filling a puzzle or categorizing whether this is a semaphore or a tractor or bus. These sort of mechanisms now can be trivially automated with AI. And so AgentKit is a tool that we built to solve this very issue. You can

now register an agent with AgentBook within AgentKit. So you can register an agent sort of in like a registrar of agents. And now you know with certainty that an agent is acting on the behalf of a real human that is verified with World ID. And as a as a as an artist you can gate or restrict sales to like only five tickets per human, right? Or per proof. So now

now you know that an agent interacting on the behalf of a human cannot attack or spam ticket sales. So now you know that the tickets actually go to real fans as opposed to bots that just try to extract money out of fans because they just want to go see a ticket.

Definitely. And that example with Ticketmaster has been an issue for decades. I built a whole company for

blockchain secure ticketing back in the 2017 days to try and solve it and that ticket mafia, man, they're they're they're crazy. And with AI agents that it definitely makes it more complex, but do you have to In that example, would you have to force Ticketmaster to if to How do you stop the bots? You know, obviously they're the people who own the bots are not incentivized to

get a World ID cuz it limits them. So do you have to work with the partners to enforce that technology?

Yeah, totally. I mean it's it's not a question about forcing, but more so about creating a value proposition that is appealing to these businesses and services. If they feel like a lot of their customer experience is really degrading because of bots and AI, and this is

going to get worse and worse over time, there needs to be multiple solutions to this, and we are trying to offer the best one for this. And so, we do need to work with these partners to sort of help them address a lot of these issues and build native tools for their platforms. So, we partnered with Coinbase on this because Coinbase built XPRL 2, which is a an agentic

payment standard. So, now these businesses, they can also take stable coins for their payments, and they can also associate a World ID proof with their XPRL 2 payment. That's how Agent Kit is built under under the hood. And so, so yeah, it's it's working with a lot of partners to sort of address the issues that they're facing these days. It's It's try and also get them more

acquainted with the general problem space, you know, like how this works, and try and develop solutions that work for them because we've learned a lot by working with partners. There's a lot of things that we didn't, for example, fully grasp until we constantly keep iterating on our own products because now we have a way of serving proof of human, but how do we

actually make that useful in the context of existing applications out there in the world?

Mhm, definitely. So, on the Coinbase partnership, that's really where the XPRL 2 kicks in, which from what I understand, AI agents will be paying with stable coins in a protocol that makes it fast and easy, but it looks like that's not it from you

would you would know best. Is that really not being utilized yet? I feel like most people that have AI agents, they're using it to do tasks, but it's not necessarily doing payments just yet in the mainstream for the early adopters of AI.

Yeah, totally, totally. It a very new thing. It's It's It's now a thing in the early adopter phase. Like people who are

technically able or who are who are like native in crypto, they're starting to use it. But a lot of experiences are still like traditional like card or bank-driven payments. But if you look at companies like Stripe, which I think processes something like 2% of the world economy, like it's in the trillions of dollars that of value that processes, they also

have startups like or like products like Tempo, which do the same thing. So there's a there's an analogous for XPRL 2 called machine payments protocol MPP. And so even a big business like Stripe, which processes a lot of payments, is moving into this agentic commerce world. And I think it's it's It's a race, right? Like everyone is sort of moving into this world. And as

the technology gets better, as the products get better, I think they're going to integrate with more and more partners, more and more businesses. And it's going to be eventually pushed to the end customer because eventually banks maybe will just run on crypto rails, right? Like instead of running on old COBOL or old Swift or old payment protocols, they're

eventually going to modernize as as now we have like better regulation, better systems, better products, better enterprise support. You know, like a lot of these things are eventually going to move to the new rails. And so I think give it like two to three years, I think most of the payments on the internet are going to run on these rails, even maybe without people knowing

it, right? You're just going to see pay with my dollar wallet and it's just going to be their bank, but the bank maybe uses agentic payment rails for this.

Mhm. That's great. What kind of catalyst moment do you predict or expect for more people to jump into AI agents and then to start using stablecoins as payments? I feel like right now, you know, people

There are a lot of people are using, you know, chat GPT, for example, but they don't really have a full agentic model running. Is it is something as simple as chat GPT and Claude develop more an agent tab that can do stuff and then X402 is integrated into that?

I think it's definitely towards that direction. I think I think the computer use and browser use of these

agents is not as polished just yet. It's definitely a lot better than it was 2 years ago or a year ago, but I think it's it's still much better and more familiar for people to use a website. So it's going to be a phase, right? Like there's going to be early adopters and once the models are good enough at using things like Booking or Travala or like, I don't know, like DoorDash, Uber,

ordering things like travel for you and people get comfortable with these agentic flows, there's going to be more and more demand for integrating payment options that are better, right? Instead of waiting for your bank to or like to connect a card to your your agent, it's going to be much simpler to have like a wallet connected to it and you can pay immediately and settle

immediately and not have the overhead of payments. Though I think like traditional financial services are still going to adapt to this agentic world nonetheless because the economic activity is just going to be there. But I think eventually the crypto rails are are going to prevail because they just provide a better payment experience overall.



Definitely. I'm I'm excited to see that. You know, I would love more payments in stable coins and faster and cheaper. That was the original goal of just Bitcoin coming out and then moving to stable coins is just to not have to, especially if there's foreign fees or booking, traveling in other countries.

International currency conversions.

Yeah, all these issues that AI agents

need to solve for me, you know, can my agent.

Exactly.

[laughter]

So, with the agent kit, for those people who have AI agents now, whether they're they have Hermes or, you know, open claw, things like this, is that available for them to use today or what's the status of it?

Yeah, totally. So, any agent that you give access to a wallet that has XPRT

enabled can use any service that ASK accepts XPRT as a payment. And yeah, you can you can set it up. And use it. Yes.

That's very exciting. And if you were to predict the timeline for, you know, a huge bump in AI agent use and agent kit, you know, what how quickly do you think that we'll see a huge jump in, you know, many percentages of everyday AI

users making it easy enough for them to have AI agents to utilize this technology?

I think within a year a lot of this will be much more normal than it is now. Models are getting really smart. I think I mostly interact with models on the programming side of things as a developer. For myself, I use Codex as my daily driver. And there's there's

a lot of like sort of more customer-facing products that are coming out of the AI labs. Be that Anthropic, OpenAI, or Google. It's just going to take time to push it to these products, right? So, and building product experiences that have the same level of quality and user experience polish that all the others that are not driven by agents. I think

if you give it 12 months, you're going to see a lot more of this. By the time it becomes like really mainstream, it's probably a little bit more, like 2 to 3 years until like it's it's the prevailing way of paying and using these applications. A lot of my applications are just moving to chat boxes or voice boxes. I don't code anymore these days. I very rarely. I just talk to a

chat box, right? And have it do things for me. As these tools get better, it's going to use the browser for me. It's going to interact with all the services I'm familiar with and you're going to move away from using your browser yourself with your hands and just talking to an agent and seeing things pop up on your different devices to approve or to change things.

But it's going to take a lot of time for these products to come out I think and to have good user experience.

I'm looking forward to it. And you did mention earlier the link to find where the nearest world orbs are. I can put that in the show notes below. If you want to learn more about the agent kit and Coinbase partnership, what's the best way to learn more and get started with

your agent?

So agent kit has its own documentation. I think that it's in our own docs page. Let me just quickly confirm.

[snorts]

Yes, so if you go to docs.world.org, you have an agents page. So if you're a developer, you can integrate agent kit on there on the docs. You just click on the agents page. And if you want to learn more, you can just search on

Google world agent kit and it can take you to our blog post which sort of tells you all of the partnerships announcements, how to use it, how to integrate it, and all that all that good stuff.

Sounds great. I appreciate your insights DC building into where the world is moving with AI agents and making sure we have proof of humanity to make it make it fair and

make sure that AI agents aren't overtaking the world and stealing our all our tickets and all these other examples. It's a crazy world we live in.

Awesome. Yeah, I'm really excited about the future too. Things are changing fast but for the better.

Definitely. Thanks for the time and would love to follow up again in the near future.

Totally, totally. Looking forward to

chat.

Related coverage

More interviews

Browse all 1,084 interviews

Get new interviews firstCCS Insider, the free newsletter from Ashton Addison. Twice a week.

Subscribe free