10 Minutes or Less: Vercel CEO Guillermo Rauch on AI Agents and Leadership

INTRO — Format setup (00:00)

Ali Rohde (00:00) We're just gonna cut it to the 10 most interesting ones. You're gonna have one minute to answer all of them. We have this timer. It'll run as soon as I ask the question. This is going to be super fast. We're going to have you in and out in 10-ish minutes. The goal is for this to be super interesting and high bang for your buck. I think it's criminal that every podcast is an hour and 40 minutes. Somehow everyone has coalesced around that number. This is going to be 10 minutes. It's like the lightning round at the end of an interview.

Guillermo Rauch (00:30) What the hell. Great. It's just pure signal.

Ali Rohde (00:45) Exactly. Officially, Guillermo, welcome to 10 Questions.

Guillermo Rauch (00:45) Thanks for having me. I love this format, by the way.

Ali Rohde (00:45) Thank you. We're gonna go through 10 questions, give you about a minute each. Do I have your permission to cut you off if you're going a little long?

Guillermo Rauch (01:00) Of course.

Ali Rohde (01:00) Awesome. Let's get started.

Q1 — Do AI teams need to work 80-hour weeks? (01:15)

Ali Rohde (01:15) Given the pace of AI today, do teams need to be working 80 hours per week to stay competitive?

Guillermo Rauch (01:30) My gut tells me even more so, because of the pace of innovation, acceleration, and all of the new things that you have to learn about and stay up to date on. I feel like you have to be more online than ever. I call it terminally online sometimes.

I do think you want to balance taking in a lot of input with being productive, shipping, and creating high-quality things. So it's a balance, but I don't have an exact number of hours. I work all the time because I enjoy it.

Q2 — Mustache maintenance routine (02:00)

Ali Rohde (02:00) Next question. This is submitted by a mutual portfolio founder of ours. What is the current mustache maintenance routine?

Guillermo Rauch (02:00) Funny enough, I got this question yesterday. There are many styles. Mine is recently just super clean around it, and then as thin and clean as possible. But it's funny how it's become my recognizable symbol kind of overnight.

Ali Rohde (02:30) Yes. It's your trademark.

Guillermo Rauch (02:30) It is my trademark.

Q3 — Silicon Valley from Argentina (02:30)

Ali Rohde (02:30) Growing up in Argentina, what was your conception of Silicon Valley back then?

Guillermo Rauch (02:45) It's so funny because for me, San Francisco and Silicon Valley has been almost like my true calling. You know the phrase, "people have ideas and ideas have people"? I think some cities draw people in.

I've been drawn here because I didn't intellectually know much about it. The beauty of living here is you just have to stay here, live here, and all kinds of interesting people will pass through because of the strong network effects it has.

I knew very little. I do remember a geography teacher once mentioned how due to disruptive innovation in technology and the rise of silicon semiconductors, tremendous wealth accumulation had happened here. I remember that one fact from fifth grade.

Ali Rohde (03:30) And then you're like, I must go there.

Q4 — How Argentina shaped how he leads (03:30)

Ali Rohde (03:30) What's one way growing up in Argentina, growing up outside of Silicon Valley, has shaped how you lead?

Guillermo Rauch (03:45) I don't take anything for granted. I grew up in a place where we didn't have very much. Nothing was guaranteed. This is sad to say, but not even our personal safety was something I could take for granted.

Maybe the way that I approach work, working hard, earning everything, and competing comes from that. But also, when I see something that is so much better than what I grew up with, I enjoy it so much more than the average person. I have more appreciation for things that are better and great.

Ali Rohde (04:30) Do you go back ever?

Guillermo Rauch (04:30) Every couple years. Recently, I bring my family over. Things come to San Francisco rather than me going to other places.

Q5 — What he has figured out about leverage (04:45)

Ali Rohde (04:45) You have five kids. You are, as you say, terminally online. You're always engaging with customers. You work out regularly. You're an investor in the most exciting new startups, and your team ships like crazy. What have you figured out that more founders should know?

Guillermo Rauch (05:00) That all of these things are connected. When I started at Vercel, I was not nearly in the position that I'm in today. I had very little money and resources. I should not give you a false image, because I grew up with extremely little, and once I was in San Francisco, I was much better off. But I decided I'm going to take on the challenge of creating a startup and creating a family at the same time.

I fundamentally believe that we can all take on more than we give ourselves credit for. Going back to the other question, we have it so good. Even though we work hard on all these things, we still have it so good. We live in the greatest country, the most wealthy.

All of those things you enumerated are connected. Startup investing builds a stronger world model for where technology is headed for me. So I do that to learn and to stay connected to the community and the ecosystem. It's part of the whole.

Ali Rohde (06:00) They all feed into each other.

Q6 — A hire who changed the company's trajectory (06:00)

Ali Rohde (06:00) Who is one hire that has changed the company's trajectory?

Guillermo Rauch (06:15) I have one that I was just really excited about earlier today because I was reviewing data from some performance benchmarks. We acquired a single-person company with a really exciting project called Lagon. This guy's name is Tom Leonard, and he's just an extraordinary infrastructure engineer. One of the last bastions of human capability before AI takes over everything. We still need amazing infrastructure engineers at Vercel, if you're listening. We're hiring.

I always come back to individuals with high agency, high intelligence, high integrity, who push themselves really hard. There are a lot of people at Vercel who don't wait for a PM to give them things or for customers to complain. They go out and seek info and seek alpha from the world. Obviously he's not the only one, but it's very precious because I just saw a banger performance improvement an hour ago.

Ali Rohde (07:00) Incredible.

Q7 — What he still does personally (07:15)

Ali Rohde (07:15) What's one thing you're still personally doing that other CEOs at your stage might have delegated?

Guillermo Rauch (07:15) I sadly, and I use that word purposefully, get credit for doing customer support myself. People literally say, "Oh my God, you're a CEO, what are you doing?" And I always consider, what is the alternative? What are the other CEOs doing? They don't know their product? They don't talk to customers? They think some customers are not as important as other customers?

Of course there are customers that we get more revenue from, but even the free-tier developer is informing how I can build the best platform in the world, so that then the largest companies can use and benefit from it. I do a lot of support. I talk to a lot of people every day. I read every email and DM that I get, and I use our product a lot. I think it's becoming more popular now, for what it's worth, but it still shocks some people.

Ali Rohde (08:15) I think it might be becoming more popular in part because of the example that you're setting.

Guillermo Rauch (08:15) That'd be great.

Q8 — How he uses Twitter/X and company communication (08:15)

Ali Rohde (08:15) How many times do you check Twitter a day?

Guillermo Rauch (08:30) It's hard to say. I'll tell you my framework. I still have meetings. I still do traditional company running, like weekly business review. I don't do as many one-on-ones, and I did this before it was cool. Jensen Huang doesn't do one-on-ones, but whatever. I love Jensen.

I was also always against one-to-one information sharing when I could disseminate information more efficiently with a larger group. So I do a lot more long-form writing. I write threads, bring customer feedback to the team, and publish it to bigger groups.

Then I set up unstructured time when I can go on Twitter, play with our products, do customer support, or do whatever I feel is productive at that time. To your first question, it helps me navigate this rapid, fast-moving world of AI because in those unstructured time slots, I learn about new things.

Ali Rohde (09:30) So less one-on-one time, more writing, more Twitter, more group conversation. Some people say OpenAI runs on Twitter. Would you say Vercel runs on Twitter?

Guillermo Rauch (09:45) When there was a blog post that came out about how they run their company, I remember thinking we both, maybe because we're both based in San Francisco and we like the same things, independently arrived at a very similar set of conclusions.

We do a lot of work on Slack. We've hired people from Google and Stripe and they always get culture shock because they used to work with emails and documents. I've always been a real-time communication guy. One of my earliest claims to fame was creating a real-time communication framework for building chat applications.

I've always believed in the power of chat and text. So the company runs on Slack, especially now with agents. We have an agent that you can ask any sort of business intelligence question to. Even more so now, I'm more bullish on Slack. I see X and Slack as almost a continuum because X at its best is a broader, bigger conversation with the whole customer base and potential customers.

Ali Rohde (10:45) I saw the Lenny's Podcast episode with your COO talking about your agent setup on the biz-ops side of things, which seemed incredible.

Guillermo Rauch (11:00) Yes, basically for making go-to-market a lot more efficient.

Ali Rohde (11:00) What has it been like for Jeanne to come aboard?

Guillermo Rauch (11:15) Jeanne came here with a very strong point of view that the future of software is going to be through consumption business models: you pay for what you use, growth and incentive alignment. If you're more successful with Vercel, we're more successful and vice versa.

Because of the Stripe model of attracting developers and convincing them because it's a better product, not through aggressive marketing campaigns and airport ads, which eventually they ended up doing, you start by building the best product. That should give you tremendous go-to-market efficiency.

She didn't want to build a sales org that's reliant on hundreds of bodies smiling and dialing. Agents can do a lot of that work.

Ali Rohde (12:00) It's incredible what she and you have built because you are also moving up to the enterprise, but without switching to this huge AE group.

Guillermo Rauch (12:15) That's right. We now run three of the top 25 websites on the planet. I just saw an expansion coming in from one of the largest enterprises in the world. All of this is happening with the same product-development philosophy we've always had, which is just build the best product for the end user, and then figure out how to speak to each buyer on their own altitude and language. That's part of how we're layering enterprise sales into our very strong product-led growth.

Q9 — Whether founder engagement is table stakes (12:45)

Ali Rohde (12:45) One last question on the Twitter thing. You so clearly understand the attention game. You're so good at jumping in on trending topics and engaging in thoughtful controversy. I liked your back-and-forth with the Cloudflare folks in particular. Do you think that engagement is becoming table stakes for founders, and especially DevTools founders, today?

Guillermo Rauch (13:15) Twitter aside, I've always believed that the world rewards authenticity. You can tell when someone is speaking from the heart, genuinely saying what they truly believe. I think people have a natural aversion to a manufactured message with a bunch of hashtags. Hashtags are a clear no-no. If I were doing X 101, no hashtags. It's not that hashtags are bad. It's that it doesn't feel like a purposeful message. You're just trying to exploit X as a marketing channel.

People react very negatively to that. But I think this has been true for any channel. The more that I can post what's truly on my mind, the more people reward it. To your point, that might mean navigating the current thing because people are latching on to, "Okay, the thing this week is the lady from SF that is representing China. Maybe I'll weigh in on that." Just kidding.

Typically, I like to understand the zeitgeist of technology. I also really appreciate memes. If I can find that cool intersection, I will go there.

Q10 — Metrics he tracks obsessively (14:30)

Ali Rohde (14:30) What's one company metric you track obsessively these days?

Guillermo Rauch (14:30) We have three really simple top-level metrics. The first is how much compute is under management: how many cycles of compute we enable. Vercel has always believed that what's magic about the web is rooted in computation. It's not about a static web. It's about giving you what you want at the time that you need it.

So compute under management. Tokens under management. We grew up in the era of CDNs and making the web really fast. That's content delivery network. We're now thinking about context and token delivery networks. Imagine when we have robots all around us: they'll be talking to the cloud to get information and intelligence, and there will be streaming tokens. So we measure how many tokens are flowing through our platform.

Then requests, which we're doing trillions of requests a week. This is how we measure the growth of our more classic pixels delivered over our CDN and global network.

Q11 — Whether infrastructure companies are insulated from the SaaSpocalypse (15:30)

Ali Rohde (15:30) Everyone's talking about this SaaSpocalypse. I'm sure people are asking you about it, which is mostly about app-layer companies, I think. Are infra-layer companies like Vercel insulated from that, or will they have to adapt as well?

Guillermo Rauch (15:45) I don't believe anybody has a promised tomorrow. This has always been true of the technology sector. Warren Buffett famously was apprehensive about investing in tech because we change technologies like we change underwear.

Even more so now than ever before, you have to adapt really quickly. Nothing lasts for longer than a week, it seems, in terms of fashion, customer adoption, and attention. Realistically, infrastructure is not something you change every day. But you need to really understand what people are building today with infrastructure, and that's changed very dramatically. It used to be pages. Now, for us, it's agents. Becoming an agent cloud is how we want to continue to serve our ecosystem and community.

I also don't believe SaaS is dead. I wrote an article the other day, to your point about navigating the current thing. My contribution is that you need to think about your interface for humans, which is your UI. That will evolve, and it will still be relevant. Then you have to think about your interface to agents. Having really good APIs, MCPs, and CLIs, all of these of course deployed on the Vercel platform, makes you relevant to agents.

Ali Rohde (17:15) So it's both agents and humans. Twice the work.

Guillermo Rauch (17:15) Agents and humans. Human-centric interfaces and agent-centric interfaces. You have to have both. If you don't have one of the two, you're probably cooked.

Q12 — Whether he wants to be an IC again (17:30)

Ali Rohde (17:30) Do you ever wish you could go back to being an IC?

Guillermo Rauch (17:30) Not really, because my mental model right now is that the IC is the agent. If you identify too strongly with a particular skill or individual contributor line of work, I have bad news for you, because an agent might do that particular thing better.

We're in the era of leverage, and we're all mini-CEOs at this point. Going back to why Tom succeeds at Vercel, he was a CEO of his own company, and now within Vercel he acts like the CEO of his own company. That gets rewarded, nourished, and incentivized. You've probably heard about companies that focus on hiring founders. We love hiring founders.

It's hard to say I'll go back to writing lines of code when in reality they can just have all this agentic leverage.

Ali Rohde (18:30) Got it. So no one's going to be an IC anymore.

Guillermo Rauch (18:45) No one is an IC anymore.

Q13 — Taking Vercel public (18:45)

Ali Rohde (18:45) Some founders in Silicon Valley don't actually want to take their companies public. Stripe being the number one example. Do you one day want to take Vercel public?

Guillermo Rauch (18:45) I don't understand why you wouldn't want to take a company public, especially when you're a company that is a public good. Ultimately, I see Vercel as an extension of the web and the internet. A lot of the things that we've put in place as a community and country, the regulations, audits, compliance, and reporting, those are great things.

So many news stories get written up. Who wouldn't want to have that? Okay, you can have bears and people that are really negative on your company. No one would want that. But the part of me that is competitive and wants to be pushed wants that too. If you're a CEO saying, "I don't want to go public because they're going to be mean to me," then you've missed the point. I thrive in customers being somewhat harsh, mean, and critical of our products.

Is every company ready to take on the full weight of everyone's opinions, longs and shorts, bears and bulls? Maybe not every company is ready for that. But certainly we want to be.

Ali Rohde (20:00) I think you being terminally online is good preparation for going public. You've had the bears for years. Welcome. Grab a chair.

Guillermo Rauch (20:15) Totally. I have them in my DMs and my replies right now. It's happening as we speak. And by the way, I do think the world is becoming more and more public and transparent. This is fundamentally inevitable. If you thought you couldn't go public, someone will create a Polymarket for your performance where the wisdom of the crowds will manifest through some kind of voting anyway.

Ali Rohde (20:45) Any thoughts on timeline?

Guillermo Rauch (20:45) Not really, other than we see ourselves already as a public company. We operate that way. I try to be careful with the things I say on podcasts for that reason.

Q14 — What he looks for in founders (21:00)

Ali Rohde (21:00) Last question. You're an extremely prolific angel investor. You have an incredible portfolio. What do you look for in founders?

Guillermo Rauch (21:00) I built theses in my mind of what the relevant things are. Obviously, I pay a lot of attention to whether the founder has an idea worth exploring, so that I can almost use an investment as running with a hypothesis in the background.

I was chatting with a colleague today. I was telling her what I love about agents is that it feels a little bit like Christmas. I throw in, I give an agent a task, it might take an hour, and then it's the joy of seeing the output of all those results. Like when you give GPT-5 Pro a research task and come back, that's so cool.

What I want to do ultimately is create a system where all of my ideas can be executed on, whether within Vercel, where we have a portfolio of bets and we bring in founders who can be the owners of those bets, or bets that are more out there that I can fulfill through investment into other companies or using agents myself. Basically, constantly trying out new things.

Ali Rohde (22:30) What strikes me when you speak is the word leverage. It feels like you're finding leverage in all of these different channels. Guillermo's vision through all these different channels.

Guillermo Rauch (22:30) Correct. And this is not to Obama-congratulates-Obama pat myself on the back, but my most successful investments are things that I wanted to pursue myself and they were a competing interest. I simply do not have the hours in the day.

For example, Scale AI. I was obsessed with an idea similar to Scale AI, but I had just started at Vercel, and I was like, Vercel is my favorite idea, but I would love for somebody to do this other thing. Sometimes I make investments that are very concrete: I hope someone works in that space. And if I get lucky, and you have to get lucky, you can have your cake and eat it too.

Ali Rohde (23:15) I love it. That's it. Thank you so much.