How Paul Built Browser Base into a $27M AI Browser Infrastructure Business

Introduction

Discover how Paul, founder of Browser Base, successfully raised $27 million in just 9 months through strategic fundraising rounds. In this exclusive QnA, Paul shares his AI startup playbook, revealing how he built a browser infrastructure company that powers AI agents across hundreds of AI companies. Learn the exact strategies behind his $2 million pre-seed, $4.5 million seed, and $21 million series A funding rounds, and how he maintained focus from day one to build a company people rave about working at.

Founder Success Story QnA

You’ve raised $27 million in just 9 months across three funding rounds. What was your strategy for such rapid fundraising success?

I am exhausted, man. Just what a 2024 was – a hell of a year. This is an era of a paradigm shift. There’s a whole new wave of startups coming around building AI, which is allowing for stories like Browser Base to exist where you see pretty quick validation of ideas and then actual growth and traction. Markets are moving quickly and the rising tide lifts all boats, but there is some tactics there. For us to go from conception to series A, from $0 to $27.5 million raised all in the span of three quarters, it actually took years of thinking about startups and failing at startups and building things before I was able to get to the point where we could kind of one-shot this thing.

What was Browser Base initially when you pitched to investors at the beginning of this year?

Browser Base is a web browser for AI. We are building the web browser that AI can connect to and control to interact with the web on your behalf. Imagine you want to tell your AI to go out and book a haircut for you or help your kid apply to daycare or renew your driver’s license. That’s not going to go to an API; it’s going to go to a website using a web browser, the same one that you and I use. Browser Base runs many web browsers in the cloud. We have a framework called Stagehand and it’s an open-source framework that makes it really easy for AI to control a web browser. Our customers, which are hundreds of AI companies, use Browser Base to power their web browsing features inside of their AI products.

Most startups pivot multiple times, but you stayed focused from day one. What made you so confident in your initial vision?

It started with 3,000 words. I wrote a memo about a year ago today, in November, and I wrote a memo about this opportunity. I was trying to talk myself out of starting a company. I had just sold my last company, I’d been at the acquirer for two years, I was in the best shape of my life – that’s gone away now. So life was pretty good. Why would I throw away this good, relaxing, chill white life to go build a company? I tried to talk myself out of it. I listed out all the reasons why this would and wouldn’t work. I went into excruciating detail – literally 3,000 words. Then I published it and shared it with all my smartest friends, and there weren’t really a lot of holes. I had pretty clear conviction around the market, how to get distribution, how to build the right product, what people needed, what they were willing to pay, and really laid out everything in this foundational memo for Browser Base. That was a roadmap to go build this company, and we’ve basically been executing to that.

When you were pitching investors the first time, what did you show them beyond this memo?

For the very first round, I was lucky that I had built relationships with investors beforehand. People had known me and they’d seen what I’ve done with my last company, or they’d met me outside of it. That’s probably my biggest advice for founders: you don’t want to go into a fundraising conversation and it’s a shotgun wedding. You want these investors to have had a chance to interact with you. I’d been blogging for the past couple years, and I made sure to send my articles to people who resonated with them and get their feedback on them and kind of show them my clarity of thought. The first round was a very, very light deck that took the memo I wrote and put it into a slide form. The seed was based on the momentum of the pre-seed. The pre-seed happened in less than a week, really kind of was wrapped up on the first day to be honest. The people who wanted to invest in me wanted to invest in me before I even started the company, right? Which I’m really lucky about, and that’s not every founder’s case.

What metrics were you showing investors when raising your seed round just three months later?

Not much, you know. It was pretty unmetric driven. We had some early customers, we had a product, and a really clear product vision. I had a lot of customer interviews, a lot of evidence, but it wasn’t about revenue at that time. It was really about: do we have clear evidence that this product, the hypothesis that we had in January, was validated in April? Basically, every open question I had in January was completely validated 10x by April, and the world was kind of waking up to AI agents and waking up to this tool use and models by April. It became a lot more clear. I didn’t plan to raise a seed round three months after my pre-seed, but what happened is when you raise a pre-seed, you basically say: here’s what I want to go do, here’s my open questions, and the gating factor to the seed is that you’ve actually shown evidence that your questions that you had or your hypothesis that you had was correct.

What metrics or basis convinced Kleiner Perkins and other investors to come in for your $21 million Series A?

Between the seed and the Series A, what became clear was strong product market fit. We had hundreds of customers, we had retention data, we had larger deals that were assigned. We also had a team of people who wanted to come work with us, and we grew the team quite significantly. It was team and traction, and just a clearer story once again – more evidence around where this thing can go. The strategy that we set out to accomplish across two separate rounds was coming true. The investors – CRV, Reed Christian, Bucky Moore, Kleiner Perkins, as well as OCA Ventures – those people had known me. I had been telling them this is what I was going to go do for the entire time we were doing this company, and it was happening. The best way to get an investor to want to put capital behind you is to show them you’re going to do something, then actually do it.

What’s your philosophy around building a company that people enjoy working at and rave about online?

That tweet from Eric was entirely unprompted. When Eric joined us, he actually moved into a hostel down the street. I was like, what? He’s like, yeah, I’m just staying in this dorm two blocks away from the office so I don’t have to go home all the time. I was like, you don’t have to do that, but he wanted to. I think that’s the culture that I’ve been really proud of. We never make anybody really do anything, and I think that’s the wrong approach. Some companies are like, we’re six days a week in office, be here on Saturday. When you force people to do that, it ends up creating the wrong incentives or the wrong outcome. I want to create a culture where people want to be here a lot because they really like the work that they’re doing. The way that we do that is by giving people a lot of agency, setting reasonable expectations. Our expectations are: be here for our 10:00 a.m. meeting in the morning, and then try and stay here the rest of the day so we can all hang out together. If you have to leave at 5:00, that’s fine. We trust you.

You mentioned hiring high potential people. How do you determine if someone is high potential during the interview process?

Super good question, and I wish I had a pitch-perfect answer. There’s no way. It’s not just it. There’s something about experience. I’ve been hiring people for a little while now, and this is why experience matters. I used to write off experience earlier in my career. I said experience is BS, it’s a meritocracy, it’s all about what you do. But when you get a lot of reps on hiring people, you kind of can get a sense from how they represent themselves, the opportunities that they take in the past, the things they’re looking for in terms of their next role, that will result in them being high performance. I’ve been wrong for sure, but there’s not a magic question that will tell you if someone has high potential. It’s really about: what attitude did they bring? Do they really get what we’re doing here? Do they care about it? Do they ask the right questions about why this is going to grow? Do they think that we’re high potential too?

Do you use recruiters to find talent, and what’s been your experience with them?

We do. We’ve used a recruiter in the past, and it was good. The thing that’s changed about Browser Base is when we used recruiters, we didn’t really know what our values or culture was going to be as a company. We were only 3 months old. Now that we’re 9 months old and we have 10 people and we’ve done a lot of stuff together, for the first time ever we defined our values. I really don’t think values are something you create at inception of a company. They actually happen organically, and you define them from behaviors after some time and after a core team has been built. Now we have enough people and experience where we could really define our values and culture and write them down, and that really allows recruiters to go do the work for you and say this person’s going to be a good fit, this person’s not. We’ve used a bunch of different sources for candidates. My favorite source is Hacker News. I snipe people off of Hacker News all the time. I love hiring from Hacker News – just a fantastic pool of engineers and people who are often like undiscovered gems.

What considerations went into creating your office space in San Francisco?

We’re building an in-person company, and we plan to continue to do that indefinitely. If you’re doing that, being in person is not easy. There’s days when I wake up and I’m like, oh my God, I got to go to the office today. I got to get out of bed. I was up late last night, I want to sleep in. But we all have this commitment that we’re going to put pants on every day and go to our office and spend time together. Because we know it’s inconvenient but we still do it because we value it so highly. Why not make it a little bit nicer? We took the decision to invest in an office, and that means spending thousands of dollars on both getting a lease and hiring an interior designer. The interior designer is a friend of mine – Chloe Bosman, amazing, definitely recommend her. Having her furnish the office and then making sure that we get all the good stuff in there, stock it with snacks, all this little stuff, it works. It just makes it way more appealing to come to, as well as have customers or guests or friends over. If you’re going to go in person and you’re going to require people to do something inconvenient, the least you can do is invest in the space and make it appealing or fun to go to.

For those unfamiliar, can you explain what AI agents are in simple terms?

AI agents are just software that can use tools and decide which tools to use when and how to use them. I think it’s a pretty generic explanation, so maybe I can explain a little bit more precisely. Research agents are pieces of software that will go do some research for you. They’re going to go figure out like who’s going to win the Niners game on Thursday, and they’re going to go look up stats from ESPN. Maybe they’re going to call some APIs to get the betting lines. They’re going to use a variety of different tools to come to the right answer. Perplexity is a research agent, and one of the things that they use is they use a web browser to go out and read data from the internet. Browser Base is often that web browser for many AI companies. What we’ve seen people using Browser Base for in their own AI agents has really been a combination of research agent stuff or automation work.

What are some things you use Browser Base for internally that might replace the need for a junior intern or assistant?

My go-to-market and ops lead, Beatrice, just built this cool thing. She’s responsible for our office as well, and sometimes we want to buy stuff for the office. That had to be like DM me the Amazon link, and then I’ll buy it. So she built a Browser Base Amazon bot where you can use a Slack bot to link to Browser Base, and it will go buy the thing – whatever link you give it on Amazon. That’s internal because I don’t want you to go buy a PS5. But it’s cool. We’re always asking ourselves internally: what can we build to automate something using Browser Base? We’re trying to dog food our product as much as possible. I’ve been using it for really random stuff like PR review or checking in on PR stats. For my all-hands deck, I have to put together like how many PRs did we close, and it’s kind of painful to write a script to do that, so I use the GitHub UI to go to GitHub, go to the PRs tab, and do some math. Generally, anything I have to pull data that there’s not a clear API for, I’ve been using Browser Base to try and gather that data for me so I can update reports for investors or for our internal team.

What would it look like to build a Browser Base agent that finds people on LinkedIn and sends outreach messages?

I feel like everyone wants to do LinkedIn automations. It’s a pretty common use case, and I’ve really liked what Exa has been doing with their people search recently. Exa, we work with them sometimes, really like the characteristic search that they offer as a whole API. I would say you probably don’t need to reinvent the wheel specifically because LinkedIn is just such a complex thing to automate. They do things like they ban people’s accounts. You have to log in to search LinkedIn, and then if you search too fast, they’re going to ban you, and getting banned from LinkedIn is not ideal. If I were you and I was prospecting, what I would do is actually go to the company’s job description pages and see what they’re looking for. In our case, if we want to find people who are using headless browsers, we would go to a company’s website and see if they have any job descriptions open where they talk about Chromium or Puppeteer or Playwright, and then we know that they’re a good fit for us and we can go hit them up and say, hey, I saw you’re hiring for a Chromium engineer, have you considered Browser Base? We replaced your Chromium in-house.

How would you set up an agent to pull data from multiple platforms and create weekly reports?

I would probably use our new open-source framework Stagehand for that. It’s pretty cool. The way it works is that you can use natural language to automate any website. It’s actually one of the first SDKs where the prompting is included in the code. Your code doesn’t look like traditional code; it’s actually like Stagehand.act go to this website, Stagehand.extract this data in this format. You’re using natural language – it’s like a code natural language hybrid, which is kind of unique. You’re putting the prompts in your code, and we’ve really seen a lot of community adoption around it. That’s how I built the Amazon bot we have internally. That’s how I build a lot of my stuff. You kind of need to use AI to dynamically interact with these web pages because otherwise it’s really brittle and hard to build these automations. Stagehand is really this web agent framework that makes it really easy for you to build a feature that uses AI to automate the web. In your case, if you have to go to a website, maybe you’re doing something in Shopify and you don’t have access to the APIs, you can have Stagehand go to your Shopify control panel, Stagehand login here’s my credentials, Stagehand do extract my sales from this week. Then with that data, you can plug that into some sort of report, put it in a database, put in a Slack webhook that will send you a Slack message like here’s your stuff, then you run it every single week.

You accurately predicted the last 6 months. What do the next 6 months to a year look like for Browser Base?

I’ll stretch it out to a whole year. To me, 2025 is going to be the year of the browser. My good friends over at The Browser Company said something similar, and I really love what they’re doing there on the consumer side. It’s clear to me that models are going to continue to get better, and they’re going to be able to handle more and more complex tasks. If they’re able to handle more and more complex tasks, that means they can automate more for people. But if they’re going to automate stuff for people, they’re going to need to use the things that we use to interact with our tools, which is a web browser. I think 2025 is really going to be a breakout year for Browser Base. If you thought 2024 was big for Browser Base, just wait till next year.

Paul’s Business Stats

Paul successfully raised an impressive $27.5 million in just 9 months across three funding rounds for Browser Base, demonstrating exceptional fundraising ability in the competitive AI startup space. The company has grown from concept to serving hundreds of AI customers, achieving strong product-market fit in a remarkably short timeframe.

  • Raised $27.5 million total funding in 9 months
  • Secured $2 million pre-seed, $4.5 million seed, and $21 million Series A
  • Grew from 3 to 10+ employees in under a year
  • Established hundreds of AI company customers
  • Achieved strong product-market fit with high retention rates
Funding RoundAmountTimeline
Pre-Seed$2,000,000Month 0
Seed$4,500,000Month 3
Series A$21,000,000Month 9

Paul’s Method

Paul’s approach to building Browser Base combines strategic relationship-building with relentless focus on execution. His method for rapid growth and fundraising success centers on clear vision, investor relationships, and delivering on promises.

  • Build relationships with investors before fundraising begins
  • Create a detailed 3,000-word memo outlining the business vision and strategy
  • Set clear milestones between funding rounds to demonstrate progress
  • Deliver on promises made to investors to build trust for future rounds
  • Hire high-potential people and give them room to grow with the company

Paul’s Tools

Browser Base leverages both internal and external tools to power their operations and product development. Paul emphasizes building their own product using their platform, while also utilizing specialized tools for recruitment and office management.

  • Stagehand – Their open-source framework for AI to control web browsers
  • Hacker News – Primary source for recruiting engineering talent
  • Recruiters – Used strategically with clearly defined company values
  • Twitter/X – Platform for sharing insights and attracting talent
  • Interior Design Services – Used to create an appealing office environment

Key Notes

Paul’s journey with Browser Base offers several critical insights for founders looking to replicate his success in the AI startup space. These key takeaways highlight the intersection of vision, execution, and relationship-building.

  • Having extreme clarity in what you want to build from day one is a superpower
  • Build excitement around your company before officially fundraising
  • Each funding round should strategically position you for the next round
  • Company values should emerge organically from behavior, not be created at inception
  • Investing in physical office space pays dividends in team cohesion and satisfaction

Get Started in Just 5 Steps

For founders looking to replicate Paul’s success with Browser Base, here’s a five-step approach to building an AI startup that attracts rapid funding and achieves product-market fit.

  • Write a detailed memo outlining your business concept, market opportunity, and strategy before starting
  • Build relationships with potential investors by sharing your ideas and insights through blogging or content creation
  • Create a minimum viable product quickly to validate your core hypothesis with early customers
  • Set clear milestones between funding rounds and deliver on promises to build investor trust
  • Hire high-potential people who believe in your vision and give them room to grow with the company

Conclusion

Paul’s journey with Browser Base demonstrates how strategic planning, relationship-building, and relentless execution can lead to extraordinary fundraising success in the AI startup space. By maintaining extreme clarity of vision from day one, building investor relationships before fundraising, and delivering on promises, Browser Base achieved what many startups only dream of – $27.5 million raised in just 9 months and strong product-market fit. As we move into what Paul predicts will be “the year of the browser” in 2025, Browser Base is well-positioned to capitalize on the growing need for browser infrastructure that powers AI agents. For founders looking to replicate this success, the key lessons are clear: start with vision, build relationships, execute relentlessly, and always deliver on your promises.