The Next Trillion Users

Opening

• Greeting; “first fifteen, you have the fika” joke
• Returning to NDD: from the Internetdagarna side-track at the other Clarion to “my favorite date on the calendar” (minus the two years that shall not be mentioned)
• Thanks to LG and the organizers
• Frame the talk: we’re going to talk about domains and AI

Hook — “Raise your hand if you’re a real human being”

• In this room, still 100% human
• Online, that stopped being true a long time ago — bots, crawlers, etc.
• But now there are new digital entities

How we got to agents — a short history

• ~3.5 years ago we started chatting with computers — new mode, but bounded
• We ask, they answer; if we go silent, they go silent; they can’t do anything
• Add context: folders, files, spreadsheets — “projects”
• Add tools: email, calendar, project management, Google Docs
• We become the bottleneck — only so many conversations at once, also have humans in our lives
• Add a goal + a check + a loop → that’s an agent

What an agent actually is

• Prompt to an LLM + tools + goal + ability to check + ability to loop
• Not complicated — no coding required, just talk to the LLM
• Encouragement: try it

Not smart, just relentless

• Sorry to spoil it: AI agents are not actually intelligent
• What they have that I don’t: infinite energy, discipline, persistence
• The collaboration works because they bring the relentlessness, I bring the intelligence
• They cheerfully chew through what would be tedious for me

“Book me a restaurant for tomorrow night”

• One sentence is enough
• Memory file already knows: gluten-free
• Calendar already knows: Stockholm, at The Clarion, until Wednesday
• Send it, move on, check the result later

The scaling unlock

• Many agents in parallel removes me as the bottleneck
• An agent can even coordinate other agents — I don’t have to be in every loop

Meet Totoro

• First personal agent, installed February
• Named for the Studio Ghibli character — tattoo on my ankle for ten years, not a trend-hop
• Totoro: discreet, friendly, appears when you need help, disappears when you don’t

The A-Team — Totoro promoted to Chief of Staff

• ~1 month ago: 8 specialist agents now report to Totoro
• Examples:
  • Meeting recorder → extracts action items
  • Linear manager → surfaces today’s tasks, low-hanging fruit
  • Proofreader / copywriter for blog drafts
  • Web researcher (kept separate so it doesn’t pollute other contexts)

Why a team, not one big agent

• It’s not raw compute — I’m using ~0.1% of the server
• The problem is context collapse: morning calendar brief, news brief, meeting summaries, blog drafts… by evening the responses get generic
• Foundation LLMs are generic; context is what makes them specific
• Multiple agents = each one stays in its lane, each conversation stays sharp

How my agents talk to each other

• Telegram — already use it with family, no new tool to adopt
• DM per agent + a shared group channel
• Could hand off behind my back, but I want to see the flow

This is coming for everyone

• Soon every human will have many agents
• Personal vs. professional split → easily 20 each
• Some persistent (Totoro, the team)
• Some ephemeral (“I’m going to Sevilla for ICANN in June, spin up a restaurant-booking agent for the week”)
• Billions of people × hundreds or thousands of agents over a lifetime → the next trillion internet users
• Spinning up an agent today is mildly fiddly at the start; in months it will be like sending a text

Agents need addresses

• They must be located, identified, addressed, ownership-provable
• Emails from “you” now might be from one of your agents — recipients need to know which agent, and that it is an agent (in case it says something out of bounds)

The LG-lunch example

• “Totoro, go talk to Nova (LG’s agent), find times and places we both like, come back with options”
• Today: my agent has no reliable way to find LG’s agent
• We will need a way for agents to connect to each other without ambiguity

Pin this thought — switching gears

• Hold onto the identity question; we’ll come back at the end

The numbers

• ~8.3B people on the planet
• ~6.1B online
• ~392M domains registered (end of Q1)
• Personal estimate: ~1.8 domains per registrant (so worst case, 1:1)
• That’s ~3.5% market penetration — in any other industry, this would be called a failure
• And it’s been growing for 40 years

Round 1 — Websites

• Building sites since 1994; Netscape would auto-append .com — type a word, land on the site
• For a glorious moment, addresses showed up on business cards, vans, shop windows
• Paris, early 2000s — walking with my newborn daughter, see a painter’s van: “Tapez peintre en bâtiment dans Google, puis cliquez sur le premier lien.” (Type “building painter” in Google, then click the first link.)
• I laughed. I should have been crying.
• That was the sign domains had lost the web — typing-a-word-plus-.com was already over
• Companies like Google and Facebook had every interest in us not finding things directly

Round 2 — Email

• Same address since 1997. Backend changed ~12 times — SMTP servers, providers, mail servers in my living room — nobody outside ever had to know
• Email would have been the perfect use of domains
• Today: ~50% of email opens on iCloud + Gmail (rough; numbers aren’t published)
  • Apple’s privacy-prefetch inflates iCloud’s reported share (65–70% alone)
  • Even adjusted, it’s enormous
• “Email is an open standard” — not really. Google and Apple decide who receives your mail.
• They offer custom domains, but have no real interest in promoting them
• We lost email too

Round 3 — The battle for identity

• This one is different from web and email
• Email: Gmail address still maps to a human you know wrote it
• Websites: relatively few, you learn to navigate phishing, you pay attention when you pay
• Now picture 6.1B humans, each with agents roaming the internet, with no way to identify them — “tricky” is too weak a word

This time is different

• Do we want openai.com/paolobelcastro, anthropic.com/paolobelcastro?
• Or should we all just be named Claude?
• We need identity.

Web3 raised its hand

• Blockchain / decentralized identity works as technology
• Two caveats:
  1. Still not proven at scale
  2. Total decentralization is dangerous

Decentralization is dangerous

• Sounds beautiful in theory; in practice:
  • People lose their Bitcoin wallets because they lose the passphrase
  • People get kidnapped because someone knows they hold crypto
  • It’s money under the mattress
• And that’s just money. For identity it’s worse — lose your passport, the Italian embassy can fix it. Lose your wallet, it’s a lot harder.

DNS already works

• DNS offers the same feature set blockchain promotes — minus the decentralization
• 40+ years old, scalable, secure, fit for purpose
• Possible tweak: domains used as identity should be guaranteed never to expire — your identity shouldn’t expire
• Don’t repeat past mistakes

Starbucks thought experiment

• “Who here has a website?” — a couple of hands, plus the guy in the back: “Does Instagram count?”
• “Who has email?” — every hand goes up

The twenty-five-year mistake

• We taught people that domains were mostly for websites
• They believed us. They don’t need websites. They need email — and they’re happy without a domain
• One company’s mistake became an industry’s mistake, at scale

The new opportunity

• The agentic era — whatever exact shape it takes — will need every human to identify the digital tools acting on their behalf
• Every human will need at least one domain → subdomains for agents
• Some persistent, some temporary; sometimes multiple domains to separate contexts

Someone will own this layer

• Someone will control the agent-identity layer
• Question: a handful of platforms (Anthropic, OpenAI — same story, new logos)?
• Or us — the people who’ve been building the domain industry for 25 years?

Closing — Third time’s the charm

• We lost the web. We lost email. We cannot afford to lose identity.
• One domain in the hands of every human has been this industry’s dream for as long as it has existed
• AI agents finally give us a reason — and a way in — to persuade people
• Is it going to be us, or are we going to make the same mistake again?

Thank you

[00:00:00] Robert: This segment, we will be talking about the next trillion users, and I didn’t know we had that many people on Earth. So give a warm round of applause to Paolo Belcastro of Automatic to tell us about the next trillion.

[00:00:13] Paolo: Thank you. Thank you very much. Hey, good afternoon. yes, Oh, first fifteen, you have the fika for the afternoon.

That’s coffee and everything. All right. yes, this is the next trillion internet users. I didn’t say next trillion people. but first of all, I want to say I’m really happy to be here at, Nordic Domain Days. I remember coming the first time when it was still like a side track of, Internetdegraden.

I don’t know if I pronounce that correctly, but, it was at the other Clarion and, But it was already very warm and welcoming, and it has since become my favorite date on the calendar, except for the two years that one shall not mention according to LG. but so I’m really happy and honored to be here.

[00:01:00] Thank you for, the invitation. Thank you to LG and the organizers. Before we dive in, we’re gonna talk about domains. We’re gonna talk about AI, but I would like to ask for a quick show of hands in this room. Raise your hand, If the clicker works. Raise your hand Let’s try again

[00:01:24] Robert: There we go

[00:01:25] Paolo: Raise your hand if you’re a real human being If I don’t see a hundred percent of hands raised, I’m gonna be worried. But, the reality is that right now here, this is still true. Everybody’s still a real human being. I don’t know how long that’s gonna last, but for now it’s true. Online though, that stopped being true a while ago.

Yeah, we might have a little– is– can I have 15 more minutes? Yeah. I have to click twice every time. Oh.

[00:01:53] Robert: You-

[00:01:54] Paolo: Something… Is this a mouse?

[00:01:56] Robert: Yes. So then you have to watch out for the mouse part versus the clicking part.

[00:01:59] Paolo: All [00:02:00] right. okay. It’s much- sorry. I was saying online, that has stopped being true a very long time ago.

and, there’s been, bots and crawlers and all sorts of digital tools, browsing your websites. But, now we have

new entities, new digital entities. and that started about three and a half years ago when we, started chatting with our computers. This is a little…

all right.

Something. We’re experimenting with a new format of slides, so I take responsibility for this. This is not on the team over there or Andy. This is my ideas. we started chatting with our computer, which was new. Before we would just, write code, and they would do things. We would use buttons.

But then suddenly we have, the ability to talk to them, and they respond. And what happens is that for a while, that was it. It was, not very, interactive. it was, we ask questions, they answer, we ask questions, they answer. And if we don’t say [00:03:00] anything, they don’t say anything. And, if we ask questions, they answer.

They don’t do anything. That lasted for a moment. And then we realized that saying everything every time we talk to them was long and tedious because, it’s a lot of words that we had to type or even record. And so we started, understanding at all. We… I can provide context to those tools.

I can give, access to a folder with files and images and spreadsheets and whatever. we call those projects, or we gave them different names. We just expanded the conversation. The fundamental principle wasn’t changing. We talk, they respond. If we stay silent, they stay silent. And if we talk, then ask to do something, they just say, “Hey, here’s the output.

Copy and paste it wherever you need because I cannot do anything.” So we said, “Okay, let’s give them tools. Maybe they can get access to our, email address, maybe to our calendar, maybe to our project management or Google Docs, whatever.” and of course, we are… everybody’s not [00:04:00] that reckless.

I am. But, we gave them tools, and so now it was a step forward. We were still in that situation where if we don’t talk, they don’t do anything, but at least we had this ability to say, “Do this and that for me,” and they would give us an output. And then we figure, okay, but if we keep doing that this way, we become the bottleneck.

I can only have so many conversations at once. and I also have humans in my life, that limits my time. So we figure, okay, why don’t we figure out how to give them a goal, and then they have the context, and they have the tools, and they can try to make a step forward towards the goal and then check if they have succeeded.

If they have not succeeded, they loop back, and they do another step, and they check again and again. And so in a loop like that, they can actually achieve more complex goals without requiring our permanent supervision And

fundamentally, that’s what an agent is, right? An agent is a prompt [00:05:00] given to a large language model with a set of tools, with a goal, a capacity to check if they reach the goal, and to look back if they haven’t and make another step forward.

So agent might sound like some complicated thing, but really it’s not. And if you have never tried to set up an agent, I engage you to try because it’s– doesn’t require coding, it doesn’t require programming. It only requires to talk to your, LLM. Now, the thing is we say AI, artificial intelligence, agents, they’re intelligent.

Sorry to spoil it, but they’re not. They’re not smart, actually, but they are relentless. And so it’s interesting with the collaboration with agents, that’s what I really like is that they have what I don’t have, which is infinite energy, discipline, persistence, but they don’t have the intelligence that hopefully I can bring.

And so together we actually work, really well. And, [00:06:00] they can actually achieve goals that would be extremely tedious for me, but for them, they don’t care. They don’t experience that. and so I can say things like I can tell my agent, “Hey, book me a restaurant for tomorrow night.” Now, the beauty of this is that I don’t need to add any more details.

I don’t need gluten, but my agent knows that because we have had that conversation in the past. It’s written in its memory file, which is a text file with just context. I don’t need to say in Stockholm because my calendar says that I’m here until Wednesday. So if I say book for tomorrow night, it knows it’s in Stockholm, knows I’m at The Clarion, that’s again in my calendar.

And so it does now a level of autonomy that while not being out of this world allows me to just say one sentence and move on and do something else and then come check the result a bit later. And, that removes the bottleneck, AKA me, from the picture because now I can [00:07:00] send many agents in parallel, do things, and, so I am not involved in every one of the steps they take all the time And I can even get an agent to actually coordinate other agents.

So I don’t even have to be involved individually with each one of them. They can do a bunch of the stuff on their own and then report to me. And so all these ideas

I put in practice since about February when I installed and configured my first personal agent that I named Totoro from the, Studio Ghibli, and I have to say that was not a trendy name because of the Studio Ghibli wave of last year.

I actually have Totoro tattooed on my ankle for ten years, I can, I’m really happy because Totoro is this, very discreet, friendly little mon– big monster that comes when you need, help, but then disappears and goes away, disappears when you don’t need it. And so I liked it so much [00:08:00] that about a month ago, it got promoted.

It became chief of staff. And now I have eight agents that are specialists, each one in one area, and Totoro is coordinating them. So there’s, an agent, for example, will parse all the recordings of my meetings and then extract action items for me. another agent will be managing my linear, and so identifying the tasks that I need to do today, the ones that are low-hanging fruit and could be just a five-minute, and closing.

the ones it can help me with, maybe write a draft or prepare something. Another one, helps me with the proofreading and copywriting whenever I’m writing a blog post. I have one that I send out to the web when I need to research a subject because, I like to have a specialized agent for that, because oftentimes it brings back a lot of, brings back a lot of context, a lot of which is not super relevant to my other conversations.

So that keeps it, narrow. And, and yeah, there’s– I’m not gonna list them all, but there’s eight of them and a [00:09:00] chief of staff. And you’re gonna say Okay, Paolo, is your life so busy that you need a team of nine to handle it? And the answer is no, I don’t. My life is not that complicated. The problem is that we go back to this idea that an agent is context, tools, and a goal.

And the context becomes very confusing very fast if you talk about everything and anything during a day. So in terms of sheer power, I was probably using zero point one percent of the processing power of the server on which Totoro was running. That’s not the problem. The problem is that we start the morning with a calendar brief to highlight the important events of the day.

It would follow up with a daily news brief about the world. Then it would follow up with summaries of a few meetings, and then we would read a couple of blog posts I was drafting. By the evening, it was non– it, [00:10:00] it– not even nonsense. It was very generic, response ’cause we have to remember at the foundation, there is this large language model, and there aren’t so many of those.

They’re all the same. They’re all trained on the same data. They can do kind of anything but are also very generic. What makes them specific and pertinent is the context you give to them, and if the context is very confusing and very diverse, then they lose all the specificity. So having all these agents is fundamentally a system to keep them in their lane and make sure that each conversation I can have with one of them is, very accurate and specific.

Now, they’re a team, and so they can talk to each other. So I can talk to each one of them in– I use Telegram, but, you can use any app. That’s just because that’s the one I use with my family. And, I like using Telegram because it doesn’t require me to go to yet another tool. That’s one I already have, with family and friends.

I have a DM with each one of [00:11:00] them, a direct message, but I also have them all in one channel. So when they need to hand over something, they could technically do it in the back end, like behind my back on the server, but I like to monitor and see what’s, flowing. So most of the time they follow instruction.

Not always, but, And, yeah. So what I– the point I want to make here is that soon every human will have many agents because this team, for example, is helping me in a kind of generic, half professional, half personal way. Tomorrow, I might want to split those two out and say, “No. Here’s a team that is only on the professional s-side.

It has some accesses that the others don’t. This team is only for friends and family and my personal life.” And now suddenly here I am with twenty agents push that forward, consider that some of those agents will be persistent like those, some will be ephemeral. I could say, oh, we’re going to Sevilla for ICANN in June.

I’m gonna spin out an agent to book [00:12:00] restaurants every night. That agent will disappear after that, but he’s gonna be there for a while. And so when I talk about a trillion internet user is because I’m thinking about, hey, billions of people will have hundreds or thousands of agents over their lifetime.

Spinning an agent right now is still a little bit of a– there’s a couple of tasks that at the beginning that are not super easy because you have to get to the point where you can talk on Telegram, and then it becomes easy. But we all know that within months it’s gonna become like sending a text message.

And the agents need to be located, identified, addressed. If you have a bunch of agents acting on your behalf, this is not exactly before you would send out an email, it comes from you directly. You are the one sending it. But now you’re gonna have an email sent by one of your agents.

While it’s gonna be important to know that it’s your agent and not someone else’s, maybe also important to know it’s an agent and not directly you, just in case it says something, out of bounds. And so it’s gonna be really [00:13:00] important to find them, identify them, prove ownership, and, And they have to find each other.

As an example, I was thinking, “Hey, imagine I want to have lunch with, LG in Sevilla.” I can tell Totoro, “Go talk to Nova, his agent, and figure out a couple of times where we are both available, couple of places that we both like, and then you tell us, and we’ll pick one.” But now my agent has no way to locate LG’s agent right now.

Like,

how does that work? So we’ll have, we will have a, we will need a way for those agents to actually connect to each other without any doubt. Now, I want you to hold this thought for a second. We’re gonna switch gears. We’re gonna talk about something else for a moment. We’ll come back to that, at the end.

But I want to talk about

domain names for a moment because this is Nordic domain name, not Nordic AI agents. So today there’s eight point three billion people on this planet, roughly, [00:14:00] and six point one billion of those are online. And I don’t… these are rough numbers, plus or minus a hundred million.

And there’s three hundred and ninety-two million domains registered, I think, as of the end of the first quarter of this year. And I have my personal estimation that there are roughly one point eight domains per registrant, which is coming out of nowhere. But, worst case scenario, we can say one domain per registrant.

It cannot be less. So that’s a very low market penetration. If you go in any industry and tell them, “Hey, our market penetration is three point five percent,” they’re gonna be like, “Okay, when do you shut down? what is the… this has failed, obviously. this is…” Oh, by the way, and it’s been growing for forty years.

and so I want to look at how we got here Round one is websites. So I started building websites in nineteen ninety-four, and there were a few glorious years [00:15:00] where you would, type a word, and, Netscape back then introduced that, was just adding .com at the end. And so you would just go to a website.

That was amazing. people would just learn that, oh, you type a word. And people did learn that. The only problem is that there were like hundred thousand people online, so it wasn’t like a large share of the world population. And, but you started seeing website addresses like maybe, on a business card, on the side of a truck, on a front window of a shop.

It– there was something. It was starting. And then I’m in Paris. I lived in Paris early aughts, and I was walking with my daughter, just like newborn, and I see a little van in the street. And on the side of the van it says this

Tapez peintre en bâtiment dans Google, puis cliquez sur le premier lien. So for the non-French speaker [00:16:00] here, this means type building painter in Google, then click on the first link. I burst out laughing. I was like, “What idiots are these people? They don’t realize that if tomorrow their competitor takes the first spot, now they’re advertising for their competitor.”

It would seem so, so absurd. What I didn’t realize that I should have been crying, not laughing, because that was the sign that domains had lost the web. People already back then had switched to, I go to that place, I type, and stuff happens. So the whole thing that I type, a word or two words and .com gets added and I get to the site, over.

And over, led by companies like Google or Facebook with a very strong vested interest in you not finding stuff directly, you going to them Round two was email. email is interesting because I have had this address since nineteen [00:17:00] ninety-seven, and I’m very happy with it because since nineteen ninety-seven, I maybe– I probably changed like twelve times the platform, the SMTP servers, the system I use in the…

I had mail servers in my living room. I used a different company, another one, but nobody ever had to know. All the people that have been writing to this address since nineteen ninety-seven found me. They don’t need to know what happens in the backend. So email would have been a very good use of domains.

And yet today, this is a really rough estimation because those numbers are not published anywhere. But today, fifty percent of email opens happen on iCloud and Gmail. The actual number is much higher, but it’s an artifact of Apple because of their privacy system opening every single email from the backend and then only showing users what they want to see.

So if you look at the official number, iCloud alone is sixty-five to seventy percent. But then when you bring [00:18:00] them, to reality, that’s still ginormous. Like we think, oh, email is this open standard. No, it isn’t. Google and Apple decide who receives your emails, period. So Google and Apple, although they offer you to have a custom domain, they don’t have a strong interest in you having a custom domain.

They want you to use their domain. That’s how it spreads. And so we lost the battle of email. So there’s the question of number three,

the battle for identity. Because this is happening, and, those agents Are gonna be everywhere. And this is not exactly like website or email, because we mentioned before, email, even if I use a Gmail address, you’ll know it’s mine once I gave it to you.

You, you know that I wrote to you, and you know a human being was behind the keyboard. Websites, even easier because there aren’t that many. And yes, you can– you have to [00:19:00] navigate a little bit, phishing scammers. there are obstacles along the way, but you learn to navigate those. And again, they don’t come in such high volumes.

And generally, when you interact with a website, when you transact with a website, when there’s a potential for risk, like you’re paying, you pay a bit more attention. But if tomorrow six point one billion humans have agents roaming around the internet, if we have no ability to identify them, it’s gonna be really, tricky.

Tricky is a very weak word for that. I couldn’t find another one right now. so the question is, do we want– do we all want to have a profile that is like openai.com/paolobelcastro or anthropic.com/paolobelcastro or should we all be named Claude tomorrow? I don’t know. there’s like we need identity.

Now, there’s options out there. Web Three, all the blockchain propositions about proving ownership, proving identity in a decentralized manner. Those are things that, that work. they’ve been… [00:20:00] The technology has been around for years, and now there’s a lot of solution that has been proven to work.

But there’s two major caveats for me, and you tell me if you agree or disagree with me, but one is that they still have to be proven at scale. it’s still a fairly niche market. The second is that one of the key features of blockchain is decentralization And I posit that total decentralization is a very dangerous thing.

And actually, regular human beings do not want that. Because total decentralization sounds very beautiful in theory, but that’s how people lose access to a Bitcoin wallet because they can’t put their hands on their passphrase. That’s how people get kidnapped because they have access to Bitcoin and someone knows it.

That’s basically akin to putting your money under your mattress at home and hoping that no-nobody’s ever gonna come and steal it. And I’m talking only about money. The identity. if I lose [00:21:00] my passport, it’s gonna be really annoying, but I can go to the Italian embassy and get a new one. We’ll find a way to prove that I am who I am.

But if I lose access to a wallet, it’s gonna be a lot harder. So my point is, the DNS offers the exact same set of features that blockchain has been promoting minus the cen- decentralization, and I think that’s great. I think the system we have is over forty years old, is scalable, is secure, and it can totally fulfill the role we need it to fill.

There might be some tweaks that are needed along the line. For example, I believe that there should be a way to have a guarantee that a domain that is proving your identity never expires, because your identity should not expire. But I think we have to be careful not to make the same mistakes again, because It’s a risk.

[00:22:00] that is exactly why when, LG told me the session theme is the internet of the future, and I started to think, I have no idea what the internet of the future looks like. I can’t see the future. but I know the present, and I can peek into the past, and maybe we can learn something to be ready for the future.”

And there’s an exercise I encourage you to do. Maybe you can do it only mentally or you can do it for real. But go into a Starbucks and yell out loud, “Who here has a website?” You’ll see maybe a couple of hands raising, and then there’s this dude that says, “Does Instagram count?” in the back. But it’s not gonna be, like, a glorious moment.

But then if you yell, “Who has email?” Everybody’s gonna raise their hand. Everybody has email. And so we are in a position where we made a mistake at the beginning. We taught people that domains were mostly for websites. They believed us. They don’t need [00:23:00] websites. They need email, but they are very happy without a domain.

And that’s our mistake. And it’s been twenty-five years. All right. Domains are older than that, but twenty-five years that as an industry we have made that mistake at scale. Before it was one company mistake, it became like a large community mistake. and so I strongly believe that the agentic era, whatever shape it takes in the future, which might be very different from what I’ve been showing here, which is probably already very different in many places.

But that’s gonna lead to a need for every human being to have a way to identify the tools they unleash on the digital world and that act on their behalf on the digital world, and that everybody will need at least one domain, so they can have, subdomains for all their agents, and maybe sometimes multiple domains to separate context.

And some again [00:24:00] will be persistent, some will be temporary. It doesn’t matter. It’s your subdomains. You can do whatever you want with that. And I can tell you, someone will own that layer. That’s gonna happen. there’s no question that someone will be controlling that. And so the real question is who is gonna control that?

Is that gonna be a handful of platforms like Anthropic, OpenAI, so we switch the names and the logos, but the reality remains the same? Or is that gonna be us? Is that gonna be the people in this room and outside of this room that have been building the domain industry for, twenty-five years? So in conclusion,

we lost the web, we lost email.

We cannot afford to lose identity. And putting one domain name in the hands of every single human being has been the dream of this unders– in this industry for as long as it has existed. AI agents finally give us a reason and give us an in, give [00:25:00] us a way to persuade people. And so the question I’d like to leave you with today is that gonna be us or are we going to make the same mistake again?

Thank you.