Joshua Browder is willing to stay on hold for three hours to save twenty dollars. He said so in a 2026 interview, apparently without treating this as a confession. Most people have a threshold beyond which the refund becomes less attractive than the rest of the afternoon. Browder built a company around the suspicion that businesses know precisely where that threshold sits.
DoNotPay began with parking tickets. It grew into software for subscription cancellations, refunds and other consumer disputes. Along the way, its founder branded the service a “robot lawyer,” attracted Silicon Valley investors, proposed taking artificial intelligence into court, and settled federal allegations about what his product could actually do. The teenager who wanted to challenge official paperwork eventually had some consequential paperwork of his own.
There is a useful human question underneath the machinery. How much inconvenience will a person tolerate before surrendering a small amount of money? Browder has spent much of his working life trying to change that calculation. His career also shows how readily an appealing answer can grow into a claim that needs a much harder test.
A ticket, a form, a repeat customer
Growing up in London, Browder accumulated parking fines and learned how to appeal them. He has cheerfully acknowledged that his driving left room for improvement. The important discovery came while doing the paperwork: the arguments repeated. A mistake on a ticket, confusing signage, an improperly marked space. Different motorists could find themselves explaining much the same problem.
Friends and relatives began asking for help. In his account of those early days, he was copying and pasting letters in Google Docs before automating the work. The first version of DoNotPay, launched in 2015, turned that informal service into a sequence of questions and a prepared appeal. A favour had acquired a user interface.
He entered Stanford in 2015 and studied computer science. By the following spring, his ticket program had drawn attention well beyond the people who knew him. The early mechanism was comprehensible: collect the relevant details, choose an argument and fill in a form. Its appeal depended on making an existing procedure easier to use. Nobody needed a mechanical barrister pacing a courtroom.
“This is much bigger than parking tickets, and I’m expanding,” Browder said in 2016. The sentence contains a familiar founder’s pivot, from one irritation to a much larger market. It also captures his impatience. Once a process could be broken into repeatable steps, he wanted to find the next one.
The sandwich came before the summons
Browder’s habit of building before asking permission predates DoNotPay. As a young teenager, he made an unofficial iPhone app for Pret A Manger. He liked the sandwich chain and saw an empty space in the App Store. In later interviews, he described an awkward encounter with the company over permission to use its material. The enthusiasm had arrived before the paperwork.
He had started by experimenting with a jailbroken iPhone and selling custom themes. Software offered an unusually direct transaction for a schoolboy: make something, put it in front of people, see whether they want it. There was no need to spend a decade acquiring seniority before being allowed to try. That early independence remained central to the advice he gave other would-be founders.
His family supplied a rather less ordinary background. His father is the investor and activist Bill Browder; his grandfather Felix was a mathematician; his great-grandfather Earl led the American Communist Party. The family tree accommodates several competing ideas about how the world should be arranged. Joshua chose software and, later, venture capital as his instruments.
In a joint father-and-son conversation in 2022, Bill expressed admiration for Joshua’s choice to create things. Their public work operates on very different scales, but both have made confrontation part of their professional lives. For Joshua, the recurring opponent is often an organization that can afford to spend longer on a dispute than its customer can.
When a good letter became a grand promise
The ticket tool brought recognition. MIT Technology Review included Browder in its 2017 Innovators Under 35 selection. That year, DoNotPay expanded into hundreds of other narrowly defined tasks and offered tools for others to build chatbots. His ambition was moving from helping a motorist phrase an appeal toward automating access to procedures people struggled to navigate.
Money followed. In July 2021, Browder reported a $10 million funding round valuing DoNotPay at about $210 million. Backers in that round included Andreessen Horowitz, Lux Capital, Tribe Capital, Day One Ventures and Felicis Ventures. The company that had begun by helping people resist charges was now itself a valuable business.
“Robot lawyer” was an efficient phrase. It compressed a product explanation into two words and supplied the press with an irresistible character. It also bundled very different capabilities together. Generating a useful cancellation letter, interpreting a legal problem and representing someone in a courtroom each require their own evidence. A memorable name does little to establish where one capability ends and another begins.
The gap became visible in January 2023. Browder proposed having a defendant receive AI-generated responses during a traffic hearing. He also offered $1 million for someone with a forthcoming Supreme Court case to repeat arguments supplied through AirPods. These were plans and offers, rather than demonstrations of successful courtroom representation.
He abandoned the traffic-court experiment, saying bar officials had threatened possible prosecution. He declined to identify the officials involved. The distinction matters: the cancellation happened; the account of the threats was his. A proposed public test had become an argument over whether the test could legally take place.
Criticism also came from people trying the tools themselves. Washington paralegal Kathryn Tewson publicly challenged the quality and delivery of several legal features. Browder subsequently said he would concentrate on consumer rights. He defended the company in interviews, but the debate had moved beyond whether automation was an attractive idea. People wanted to know what they were receiving.
A consumer champion meets the consumer regulator
In September 2024, the Federal Trade Commission brought allegations against DoNotPay over its claims to substitute for a human lawyer. The agency said the company had neither tested the service against that standard nor retained attorneys to evaluate the quality and accuracy of its law-related features. The complaint concerned the evidence supporting the advertising.
A final order approved in January 2025 required $193,000 in monetary relief and notices to subscribers from 2021 through 2023. It also barred claims that the service performed like a real lawyer without sufficient supporting evidence. The February announcement made the consequence public. The company’s promises now came with enforceable conditions.
For Browder, who had built a public identity around helping consumers challenge institutions, this was an uncomfortable reversal. His own company had become the subject of a consumer-protection action. That belongs in the same account as the parking-ticket invention and the investment rounds. A useful origin does not settle every question about what follows it.
“It should be a question of what’s the right outcome, of getting justice.”
Joshua Browder, 2017
The economics of the abandoned refund
Browder’s more recent explanations return to smaller, repetitive frustrations. In January 2025, he described software negotiating with utility-company representatives, sometimes putting one AI system into conversation with another. He also talked about bringing more of the company’s American products to the UK. These were his descriptions of the work and his plans for expansion.
The business argument is easy to recognize. A company collecting a small charge from many customers can accumulate a substantial sum, while each person has little reason to devote hours to recovering their share. Browder wants software to make that arithmetic less comfortable for the company. A machine can continue a mundane exchange after a person would rather make dinner.
There is a second calculation inside DoNotPay. In January 2024, it distributed more than $1 million in dividends to employees and investors. Browder described a profitable operation with a small team. Paying shareholders while remaining private gave his career a less theatrical achievement: cash leaving the company for the people who owned it. No earpiece was required.
A spare room for the next argument
Alongside DoNotPay, Browder runs Browder Capital. He left Stanford without completing his degree and became a Thiel Fellow; the network of young builders became part of his investing world. By August 2026, he was discussing both operating and investing as established parts of his working life.
His approach can be unusually personal. In 2026 interviews, he described inviting founders to live in his spare bedroom while preparing to raise institutional seed funding. In a May discussion, he described late-night meetings and asking entrepreneurs to show live revenue records. The preference is for observable behaviour: will they turn up, explain the product and produce the numbers?
There is something revealing about this second occupation. Software scales an answer across strangers. His investing method, as he describes it, brings a founder into the house. Both begin with a practical obstacle and an inclination to get involved. Yet the spare room admits what an automated form cannot provide so readily: time spent watching another person make decisions.
“Just get started,” he advised aspiring entrepreneurs in 2025. He has followed that instruction with unusual consistency, from an unauthorized sandwich app to a parking appeal service to backing other founders. The harder part has been deciding how far a promising start entitles him to go. For the customer stuck on hold, the useful measure remains wonderfully plain: did the problem get solved, and was it worth the price?
Follow the argument
Conversations, company pages and further reading.