It's going to solve all of humanity's problems. I believe that in 10 years we're going to get together to have an asado, and we're going to say, 'Man, it was really obvious that people were going to want to own their own intelligence.' And what happens when you optimize everything? Prices go down. It's very easy to grab ChatGPT Codex, throw a prompt at it and say, 'Ah, it made a thing.' Just the other day someone was asking me, 'How do you manage to drop banger tweets?'
Well, we're here with Guille, the great G from Vercel. We ask all of our guests two questions. First, who are you and what do you do, for the people out there who might not know you?
Well, I'm the CEO and founder of Vercel. I started this company, which is basically an infrastructure and tools company for developers and agents so they can deploy software. We were talking today about how the company has been around for 10 years, well, 11 years. I started it in this place, which is the Palace of Fine Arts, right? Palace of Arts, the event, the place where I had the most beautiful event I've ever been to.
Awesome. Well, here in San Francisco, San Francisco has an incredible history of, you know, the Gold Rush and all that stuff. It's a city that was created based on people coming to try to innovate, find gold or whatever, and it's a city that has reinvented itself a lot. At one point there was an earthquake and everything was destroyed, everything caught fire. Look, they relaunched it. They did the relaunch of the city and they created this building, which is the Palace of Fine Arts.
And I remember when I started the company, I lived near here, and a coworking space started here organized by the city government. And in fact, today's event is totally full circle because in this place was my first official office, which was literally me, a dog, my dog Chihuahua, and a colleague from my previous startup here in the office. In fact, one of the first investors came, the meeting was here, who today is the same guy who leads everything AI at Meta, right? And who just launched Muse, he was one of the first investors, he came here and said, 'Man, I didn't know there was a coworking space here. We're all amazed by this building, right?' But yes, that's the idea of Vercel, right? That companies from Meta, which launched Muse on Vercel, yes, yes, yes, who launched the tools to train models on Vercel, to anyone who has an idea and wants to launch software on the internet, whether it's an agent or an app, they use Vercel.
Incredible. And what is the role of Vercel in the future of technology for you?
We always had this idea of democratizing. What does that mean? Well, what I was telling you about Meta is very interesting because when we started the company, I was very inspired by companies like Google, Amazon, and Meta, where if you talked to the engineers working inside those companies, they saw the future, they had access to tools, infrastructure, ideas, innovations. Whereas if you started a startup anywhere, whether in Latin America, Europe, or the United States, you didn't have access to that hardware that those guys had. So the company started with that idea of democratizing, and I had a lot of friends and connections, a lot from the open source world. I understood it quite well, and I was always interested in reverse engineering too. And there are several of your talks in Argentina talking about... I always liked understanding how these systems work, right? So I would sit down and analyze... Just outside, I don't know if you saw her, that was our COO.
Yes, Anne. She was one of the first support engineers for Gmail.
Ah, that brings back memories. I was at Carlos Pellegrini at that time. I think I was 13 years old or around there, and I remember that Gmail launched and you needed an invitation because it was a beta and you needed an invitation to get access to Gmail. At that time I lived in Argentina, she was working inside Google, answering support tickets and creating the support infrastructure to later absorb a billion users. Now, how did Google manage to release the search engine, release Gmail? It's like they nailed them all, right? And obviously they release things that don't work and destroy them and nobody remembers, right? But they could do that because of the infrastructure. Right, right. So if you give the world really good infrastructure, there are going to be a lot of people who surely won't make it. I'm not going to sell you on, 'Ah yes, start your startup here and 100 out of 100 are going to be the next Meta,' right? But you are going to need really high-quality infrastructure that can handle it, that if a lot of traffic comes it can handle it, that doesn't break, that's secure. And now the topic of agents is very interesting because it's even harder than it was with a website to launch an agent that you can really say is yours. It's very easy to grab ChatGPT Codex, throw a prompt at it and say, 'Ah, I made an agent.' No, no, you didn't make an agent, you typed a prompt into a website. So what we're proposing to the market, whether developers, startups, and enterprises, is own your intelligence. Which is, I believe, in 10 years we're going to get together, have an asado, and we're going to say, 'Man, it was really obvious that people were going to want to own their own intelligence.' Right. And that's the bet Vercel is making.
Awesome. And on this topic of talking about the future, you recently tweeted that if technology makes everything work and agents make everything work, it starts to be, it could all be free. Yes. How does the world work in that scenario?
Look, it's very hard to predict, but it's very clear to me that even the current technology paradigm of artificial intelligence is going to solve all of humanity's problems, whether from... obviously we started with software, but one sometimes thinks, well, AI Claude Code solved coding for me, but if you think about what coding does, coding lets you solve everything. Coding lets you... you grab an agent and instead of saying 'make me a website and deploy it on Vercel,' you can say, 'solve a millennium-old math problem for me.' And it starts writing Python and starts writing LLMs and starts writing whatever and solves that problem. So there's a lot of transfer of intelligence and the current artificial intelligence paradigm to all kinds of problems. And why did I say free? Obviously I can't tell you the specific timeline, but it's like artificial intelligence is intrinsically open source. What does that mean? If I have a very good artificial intelligence, it's going to allow you to generate itself because it's going to be so intelligent that it's going to say, 'Well, how...' Even you can experiment with this today. You can tell your agent, 'Hey, help me train a model.' And maybe it won't get you to Astra. Yes, yes, yes. But that's the good thing about artificial intelligence today. You can ask it anything and it will guide you. Then maybe you reach a certain limit, but you keep pushing and that's how the different generations of models go. And obviously the economy itself, I was just talking in an interview, the economy today is agents. Today at Vercel, 60% of deployments are through agents.
