Starting a venture capital backed startup is statistically a very bad idea. With success rates in the single-digit-percentages, there are much more reliable ways to make money.
I should know. After starting, growing, and selling my own startup, I learned firsthand how hard it is to build a durable company. And after that experience, I told everyone I wouldnāt do it again.
Thatās because startups require extreme amounts of hard work, sacrifice, and resiliency. And the only way to put in this amount of effort consistently is to deeply understand the problem youāre solving and the customers youāre solving it for.
This depth of understanding is rare, and canāt be brute forced. My friend Zack Hargett calls it āearning a secret.ā Starting a company and then searching for a problem to solve almost always results in failure. Itās like trying to find a diamond by digging in the sand.
But this time, I think I found my diamond.
This memo has always been a testing ground for our ideas. As much as we appreciate our 200 subscribers (Hi Mom!), weāre really writing it for us.
And writing is thinking. In my memo, An Internet Browser for AI, I laid out my thoughts on how internet browsers are going to become a crucial piece of software infrastructure for AI applications.
I didnāt write it with a plan to start a company. I just felt passionate about the space. The type of passion that inspires 3,000 words in a single day. I've been thinking about this problem for the past four years at two separateĀ companies.Ā
After publishing that memo, I started hearing from developers who were just as passionate as I was. One of them sent me an email that was over 1000 words long! Fifty customer interviews later, and it was clear that there were many developers who were as frustrated as I was.
Thereās a special moment in a founders journey where you switch from searching for market validation to building the MVP. Youāll know youāve arrived when you start to feel anxiety that you should have started building the product weeks ago. Thatās a good sign.
For me, it hit me after a conversation in Salesforce Park with a potential customer. The one who wrote me the long email. He said something along the lines of, āIf you build this, Iād definitely pay for it.ā So, I walked back into my office and kindly gave my notice.
Hereās what pushed me over the edge:
Every developer has a painful web automation story
As I spoke to more customers, I was delighted how many unique use cases I found. Every developer has written some piece of software that attempts to automate something in their lives. Booking a haircut, checking for new posts on a website, or even fetching their most recent PG&E bill online. Building a developer tool thatās too niche can be a death sentence for adoption. But the web is universal. And developers have been hacking on it in interesting ways for a long time.
This is something that should ājust workā, but doesnāt.
Writing code that automates an interaction on a webpage is surprisingly difficult. I cover this at length in my original memo. But the especially unique insight here is that the amount of maintenance required to operate web browsers in the cloud is significantly higher than traditional pieces of developer infrastructure. Not only do developers need to use complex tools like Kubernetes, but they also have to deal with unreliable vendors (such as proxy providers) to prevent being blocked.
Demand is increasing measurably and developers are underserved
There aren't many companies who are focused solely on building reliable headless browser infra. Generally, the products in the market are often secondary products of scraping (ie:Ā BrightdataĀ or Apify). That makes sense, as the bigger opportunity has always been in web scraping vs web automation. But with the rise of AI, there is demand for an AI-first headless browser platform. We intend to serve any AI company that needs to leverage headless browsers with their own models, agents, or AI-powered workflows. And we'll do it better than any legacy provider.Ā
I was made to start this company
I started my career at Twilio, who made an API for text messages. My last startup was acquired by Mux, who made an API for video. And now Iām starting Browserbase, an API for web browsers. Clearly, I have a thing for API companies.
Following your passion means doing something that energizes you even when youāve barely slept the night before. For all of 2023, I was enjoying the work I was doing but didnāt feel like it was energizing me. With Browserbase, Iāve found more energy than Iāve had in years. And a lot less sleep.
Overall, I believe the opportunity to build modern headless browser infrastructure is so compelling that Iāve raised $1.8M in pre-seed funding from Alana Goyal, Yuri Sagalov, Guillermo Rauch and over a dozen other founders/operators in developer tools. We're well capitalized and hiring great engineers. If youāre interested, please send me a note: paul@browserbase.com