David Hsu looked out the window of his childhood bedroom in Palo Alto. It was 2017, and for the greater part of the year, the room had served double duty as Hsu’s living quarters as well as the headquarters for his peer-to-peer payments startup, Cashew. Hsu’s parents had welcomed their son and the Cashew team into their home, sharing meals, couch space and the washing machine with the 20-something-year-olds. 

A few months earlier, Cashew had been accepted to Y Combinator, but more recently, on Demo Day, when participants present their companies to potential investors, the Cashew team remained seated. They had nothing. Unable to make Cashew’s financial model work, Hsu and his team had burned through half their cash, and would ultimately decide to call it quits.

Hsu recalls the pain of this moment. “Our peers were up on stage raising all this money, and we felt like we let all the people who believed in us down,” he says. “I remember thinking, ‘What are we doing? Why are we burning away our savings, working on an app that loses money on every transaction?’” Mulling it over in his childhood bedroom, with its walls covered in punk-rock posters—relics that reflected his teenage rebelliousness and fighting spirit—Hsu decided he wasn’t done yet. 

“Failure is something you can recover from,” Hsu says. “I thought, ‘Okay, what did we learn? And how can we apply that to what’s next?’” From this soul searching, Retool was born. 

Retool is a software development platform that offers drag-and-drop, customizable building blocks for developers to quickly assemble commonly-used software systems—admin panels, support dashboards, customer portals and more—so that they can focus on core features that set their businesses apart. “Our view was that the way people were building software was ridiculously inefficient,” Hsu says. “There’s no need for everybody to be constantly writing all this boilerplate code.” While critics bristled initially at this belief and found elements of Retool’s business counterintuitive, Hsu stuck to his guns. Today, the platform launched from Hsu’s childhood bedroom powers tens of thousands of companies including Disney, Airbnb and Mercedes-Benz.

“Failure is something you can recover from. I thought, ‘Okay, what did we learn? And how can we apply that to what’s next?’”

David Hsu

When Hsu was 16, his parents told him they’d had enough. His grades were terrible and his friend group—fellow punk-rock fans who favored mosh pits and garage jam sessions over school—was, in their eyes, even worse. Hsu’s parents, who had immigrated from China to the United States for post-graduate studies, had tried everything, from curfews to taking away Hsu’s bicycle (Hsu simply walked instead). They informed their son they were moving him to Saratoga, a quaint suburb near Palo Alto. No more wayward friends, no more East Bay counterculture to distract him. 

Hsu acknowledges that, angry as he was, the change helped. With fewer diversions, he started thinking about his future. Hsu was already an avid Hacker News reader and remembers an essay where Paul Graham, the venture capitalist who co-founded Y Combinator, described the creation of startups as “almost a measurement of the accuracy of your worldview. In a startup, there’s nowhere to hide,” Hsu says. “You can’t hide behind the work of others, you can’t fake your way to long-term success. You have to work hard (since otherwise you fail), you have to be smart (since otherwise you fail), and you have to be correct often enough (since otherwise you fail). But if you do it right, the impact of your work is remarkable: you can create something that actually puts a dent in the universe.” As a rebellious teenager who dreamed of having an impact and proving to others that his worldview was correct, Hsu was enamored by this challenge.

This fascination was heightened by the fact that Hsu’s high school experience coincided with the rise of Twitter and the Arab Spring. “I watched as this website, which was created in a matter of days, just miles from where I lived, topple a dictator halfway around the world,” Hsu says. “Technology could change peoples’ lives for the better, and that was exhilarating to me.” 

The Silicon Valley founder trope would suggest Hsu might forgo or drop out of college to focus on technology. Instead, he did the opposite. He hunkered down, was admitted to Oxford to study philosophy and computer science, graduated high school early, and spent a gap year in China. “I wanted to learn more about the innards of computers,” Hsu says. “But I also wanted to leave the bubble of Palo Alto.” 

As for his interest in philosophy, Hsu says it stemmed from Douglas Hofstadter’s Gödel, Escher, Bach: An Eternal Golden Braid. He recalls a particular chapter about the relationship between ants and anthills. In it, a crab, anteater and tortoise discuss the paradox of how individual ants (who are not intelligent), when formed into a colony, can demonstrate signs of intelligence at the colony level. “This phenomenon is similar to neurons in the brain, where any individual neuron isn’t doing anything particularly smart,” Hsu says. “But then you put all the neurons together and you create a brain. That’s pretty smart.” Back in 1979, Hofstadter also used this logic of emergent intelligence to predict computers gaining the power to think like humans. “These philosophical ideas about humanity and science really interested me,” Hsu says. And then, of course, “the other reason I chose this area of study was that philosophy is kind of a big f*** you to your parents.” 

“I wanted to learn more about the innards of computers. But I also wanted to leave the bubble of Palo Alto.”

David Hsu

“What’s the worst that could happen?” 

It was 2015 and Hsu, now in his senior year at Oxford, had gone out for ice cream with classmates. “We wanted to pay each other back,” Hsu says, “But England didn’t have an easy way for people to send money to each other.” Venmo, so popular in the United States, didn’t exist in the U.K. Hsu turned to his friends: “I was like, ‘It can’t be that hard. I bet we can build something similar.’” 

One year earlier, on a whim, Hsu had joined his college’s Ball Committee, which organized the school’s seasonal dances. He quickly realized that ticketing for the balls was an unwieldy process. “Technology is unevenly distributed across the world,” Hsu says. “In the U.S., you have good ticketing platforms, but there wasn’t anything good in England.” So Hsu, having never built a production ticketing system before, decided to engineer a system where students could RSVP to the balls, buy tickets in advance, and check in by scanning their wristbands by the door. “It wasn’t particularly complicated, but attendees were shocked: there were no queues for purchasing tickets, the site handled thousands of concurrent users successfully, and scanning the wristband admitted you directly to the ball,” Hsu says. “If I had been back in Silicon Valley and built a ball ticketing site, no one would have cared. But at Oxford, where there were very few computer scientists, the skillset I had was valuable, and I was happy to have found a use for it.” 
Back at the ice cream parlor, Hsu saw a similar opportunity and laid out his argument: there was a need in the U.K. for a peer-to-peer payments system. No one had built one; why shouldn’t they? What’s the worst that could happen?

“It’s a terrible idea,” John Collison said. The Stripe co-founder sat across from Hsu in Mountain View, at Y Combinator’s office. It was the winter of 2017, and in the time since the ice cream outing, Hsu had graduated from Oxford, launched Cashew, gained a small, local following for the app, and been accepted to Y Combinator.  

Hsu had 30 minutes booked with Collison. He spent the first three minutes pitching Cashew. Collison, in turn, took the next three explaining why it wouldn’t work: Cashew lost money on every transaction with fees paid to credit card providers, and while this was negligible in the startup’s early stages, if the app’s adoption continued, the cost would become a death sentence. Building out a business model that could offset transaction costs in the U.K. would be next to impossible. “I had arguments against what he said, but he rebutted each one,” Hsu says. “He was like, ‘no, nope, no, no, no.’ After maybe 10 minutes of our 30-minute allotted time, there was nothing left to say.” 

Had Hsu taken a page from his philosophy textbook, perhaps he would’ve listened to Collison. Hsu says one of his main takeaways from studying philosophy relates to the idea that reality isn’t objective. “People see the world from different angles,” Hsu says. “The way we react to things, evaluate pros and cons, and generate opinions is based on our own axioms and lived experiences. There’s no one ‘right way’ or ‘right reality.’ And so approaching situations with empathy and humility, seeing the world through someone else’s perspective, and then often revising your previously held beliefs, is critical.”

Still fresh from uni, however, this lesson hadn’t fully registered. Hsu left his meeting with Collison (respectfully) defiant and determined to prove him wrong. “And then, of course,” Hsu says, “he was right.” 

By the spring of 2017, the company was hemorrhaging money in transaction fees and its bank account was rapidly dwindling. The Cashew team had attempted to cut its losses by convincing retailers to take payments via a Cashew digital wallet, but its development proved costly and time-intensive. “Looking back, I was incredibly naive,” Hsu says. “I had a pie-in-the-sky sort of attitude about what I thought of as this rebellious undertaking that everyone said wouldn’t work. Cashew had started as a project with friends, right? But after Y Combinator invested in us, after I moved back home, it was like, ‘Okay, this is real now. It’s time to get serious.’” 

While Cashew’s Achilles’ heel was its financial model, Hsu recognized that the internal systems the team had developed to support the app were powerful. The process of building these systems, however, had been repetitive and inefficient. “I remember always thinking, ‘I wish there was a faster way to build this software,’” Hsu says. “And so, post-Cashew, we had this idea of, ‘What if we took the code that we already had from Cashew and repurposed it? What if we gave it to other developers so they didn’t have to build the same internal systems from scratch?’” Hsu says. “This was the provenance of Retool.” 

“I remember always thinking, ‘I wish there was a faster way to build this software.’”

David Hsu

As with Cashew, Hsu’s new idea—offering visual building blocks for engineers—had its share of naysayers. “Most people thought it was a pretty bad idea,” Hsu says. “Why would engineers buy something they could build themselves? Why not target non-coders? Why give engineers a vitamin when you could give non-coders a lifesaver?” 

Companies offering low-code or no-code solutions were trending at the time, but Hsu’s take was that, in the case of what Retool was looking to solve for, targeting non-coders was unrealistic. “Yes, these are building blocks, but they’re not one size fits all. They require customization and stitching together into a system that is fairly complex,” Hsu says. “You need to be able to talk to your computer, and to code, to achieve this.” 

As for the argument of why engineers wouldn’t simply build the features Retool was offering themselves, Hsu’s view was that a company’s edge is its core product, not its internal systems. “Why not provide building blocks for those internal systems so that engineers can focus on their company’s core product, the custom stuff that’s fun, creative, exciting and differentiating?” 

While the echoes of “this won’t work” from a previously failed endeavor might have stopped someone else, Hsu’s individualistic streak won out again. But he was also no longer the “pie-in-the-sky” college graduate who pitched John Collison and plowed ahead despite his warnings. His experience with Cashew had taught him “just because you’re a contrarian, it doesn’t mean you’re right,” Hsu says. “To know whether you’re a contrarian and right, or a contrarian and wrong, you have to take the time to fully understand the world in which you’re operating, the customers you’re serving and the business you’re building.” 

To achieve this, Hsu began talking to other engineers to see if a product like Retool might be useful to them. “I remember one day going into Y Combinator in Mountain View and going around to all the other startups asking, ‘Is it just us, are we the weird ones, or would this be helpful?’ And they responded enthusiastically to the idea,” Hsu says. 

Bryan Schreier, the Sequoia Partner who led Retool’s Series A funding, was one such early enthusiast. “I used to work at Google, and I managed a large team of engineers who built internal tools. Despite being mission-critical, those engineers couldn’t wait to graduate and take on higher-value work. And when they did leave internal tools, they went on to build Gmail and Google Drive. If we had had something like Retool, we could have liberated all those engineers and the team wouldn’t have needed to exist at all,” says Schreier, “so for me, it was just clearly obvious that a company should address this space.” 

“If we had had something like Retool, we could have liberated all those engineers and the team wouldn’t have needed to exist at all, so for me, it was just clearly obvious that a company should address this space.”

Bryan Schreier

This support from the YC collective left Hsu determined to pursue Retool in earnest. “Later that day, during the 20-minute drive home to Palo Alto, I had this conviction that our contrarian view was right. That the way engineers built software was ridiculous. We could build a company to change that,” says Hsu.  

When he returned to YC for Demo Day in the summer 2017, Hsu stood confidently in front of an audience of investors. The team had already signed a large enterprise customer to a $1.5 million pilot. Redemption achieved. 

In keeping with his commitment to know his own company nuts to bolts, instead of hiring a head of sales at the onset, Hsu took on the role himself. “I’d never done sales before and always pictured that scene in Glengarry Glen Ross where it’s all about charisma and closing,” Hsu says. “In reality, sales hinges on empathy and being able to explain your product. It’s about figuring out what a win-win situation could look like for both of you to work together.” 

Each time a customer signed up for Retool, Hsu remembers “it felt unbelievable.” But these achievements were ones he and his team barely acknowledged. After the failure of Cashew, Hsu was skeptical of success. “I didn’t want to celebrate prematurely. Every customer felt like it might be our last. Eventually someone’s going to find out that we’re writing code from my parents’ house!” 

Parental HQ notwithstanding, Retool’s users grew steadily in numbers, and Hsu’s approach to his new venture reinforced just how far he had come from his Cashew days. “With Retool, we prioritized having a sustainable business from the get-go, so we cared quite a bit about unit economics and stayed very lean as a company. Compared to Cashew, where every transaction cost us money, this was a pretty different mindset. Here, people were paying us to use the software we wrote—in my childhood bedroom. Wow.”

Today, Retool has tens of thousands of customers; it’s helped businesses save hundreds of millions in operational costs; it boasts use cases including forecasting for the Olympics and operating networks of satellites. But Hsu’s focus on pragmatism with respect to his business remains strong. “Yes, Retool’s much larger now,” Hsu says. “But even today, we don’t want to oversell ourselves.” As Retool’s customer base grows, Hsu feels it’s more important than ever to remain empathetic and open-minded in order to understand their increasingly varied and complex needs. 

For example, in 2017, a customer suggested that Retool be able to connect to application programming interfaces (APIs) as opposed to just a company’s databases. “For a while, we pushed back on this,” Hsu says. “We were like, ‘We think connecting to databases alone is fine and this is a request we can ignore.’ But then we self-reflected and thought, ‘Actually, you know, why not? Is it a core belief of ours that Retool must not connect to APIs? No.’ So we changed our mind, and today APIs are our most popular resource.”

“Our approach was, rather than immediately jumping on the AI bandwagon, let’s talk to our customers and understand how we could use AI to actually deliver value to them.”

David Hsu

The ability to connect to APIs has proven fruitful in other ways, perhaps most significantly in putting the pieces in place to help Retool’s customers make use of AI—a shade of Hofstadter’s Gödel, Escher, Bach prediction realized. “There’s a lot of hype around AI, and we’ve seen a lot of startups in Silicon Valley try to rebrand themselves as AI companies,” Hsu says. “Our approach was, rather than immediately jumping on the AI bandwagon, let’s talk to our customers and understand how we could use AI to actually deliver value to them.” They found that most of their customers were interested in AI, but weren’t sure how to approach it or how it could help them. 

As a result of these conversations, Retool developed a suite of AI features that work in tandem with its current library of tools. “These features fall into two categories,” says David Dworsky, Retool’s Product Manager for AI. “The first category helps engineers build applications faster—so using AI to assist with code generation, query generation and debugging. The second category helps companies expedite AI-powered workflows that are specific to them. So if you’re an insurance company, using AI to match a claim with a policy. Or if you’re a fintech company, using AI to flag fraudulent transactions.” 

Because Retool is already connected to its users’ databases and APIs, “you can start stitching AI actions together with that data in literally a matter of minutes, without having to become an AI expert,” Dworsky says. “And we’re enabling these actions to be backed by whatever vendor a company wants to use—OpenAI, LLaMA, Cohere, etc. Our key insight here was to maintain flexibility behind the scenes, while making our AI features really intuitive to use.”  

For 16-year-old Hsu, a teenager with punk aspirations, siloed from friends and under the rule of parents, the prospect of founding a startup and getting to be his own boss was enticing. But today, at 28, Hsu recognizes that the job demands more than just rebellion. “Yes, you have to have your own differentiated view of the world and stand by it,” Hsu says. “But, when it comes to every element of your business, you have to be open-minded. You have to acknowledge when you are wrong, be willing to change your beliefs, and take a philosophical approach to understanding your customers and the world.” 

Hsu no longer works out of his childhood bedroom: Retool now operates out of offices in San Francisco, New York and London. While the Retool of today differs from the company at its outset, Hsu has made sure its humble beginnings aren’t forgotten, documenting the company’s journey via photos. “That’s a piece of advice I have for founders,” He says. “Take pictures every step of the way. Five, six years out, you can look back and see how far you’ve come.” 

In the case of Retool, the photos on its walls show a team that has expanded from a nonconforming party of just a few, to hundreds of people around the globe who recognize the value of thinking outside the box, and the possibilities that come with questioning, retooling—and occasionally rebelling against—the status quo.