By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
CHAPTERS
- 0:00 – 3:04
Intro and DoNotPay's Origin in Parking Tickets
Steph Smith frames the intersection of law and technology and introduces Joshua Browder. Browder explains how his own parking tickets led him to build an automated template tool that went viral and grew into DoNotPay.
- •Browder got about 30 parking tickets between ages 18 and 20
- •Friends and family sent him photos of their tickets, which led him to automate defenses with templates
- •The tool went globally viral, showing the idea reached beyond tickets
- •He has spent almost seven years expanding into other areas of law
- 3:04 – 5:47
Surprising Laws and Use Cases: Raffles, Egg Donors, Lost Pets
Smith asks about some unusual products on the DoNotPay site. Browder explains the free-entry competition law, egg donor rights, and lost pet and airport lost property forms, all built on obscure consumer rights and government services.
- •An obscure law lets people enter competitions for free, which DoNotPay automates
- •Egg donor rights and compensation sign-up is a controversial product
- •Lost pet and airport lost property forms are automated
- •Positioning: saving people time on obscure consumer rights and government services
- 5:47 – 8:29
General Counsel for Consumers: Refunds and Scale
Browder describes DoNotPay's scale and its most popular products, such as refunds and subscription cancellation. He explains the 'squeaky wheel' approach of aggressive template letters that often win refunds.
- •Over two million cases submitted and hundreds of thousands of active subscribers
- •Refunds, cancellations, suing robocallers and discounts are top products
- •A letter citing FTC statutes over broken in-flight Wi-Fi gets refunds 80%+ of the time
- •The company is shifting from decision-tree templates to true AI
- 8:29 – 12:27
From Templates to AI: Negotiating Medical Bills
Browder explains that templates were step two and AI now lets DoNotPay respond to companies live, enabling higher-value cases like medical bills. Smith raises the concern that bad counsel may be worse than none, and Browder explains how the No Surprises Act is used.
- •AI can respond back and forth, moving from $100 disputes to $10,000 bills
- •GPT scraped hospital price lists into a master file of 1.4 million rows
- •Bills are compared against nearby hospitals to request good-faith discounts
- •Browder argues consumer rights issues are not rocket science, while serious crimes may warrant a lawyer
- 12:27 – 14:49
Prioritizing Features and Where AI Falls Short
Smith asks how DoNotPay decides what to build next. Browder says the goal is to raise the value customers save, and that emotional cases like asylum and divorce are poorly suited to software.
- •The team tracks average yearly savings per customer and aims to 10X it
- •AI raises the value of cases the robot lawyer can pursue
- •Emotional cases such as asylum and divorce need a human touch
- 14:49 – 17:03
Corporate Pushback and the Bot Arms Race
Smith asks how corporations have responded. Browder describes an arms race with anti-bot systems and company AI chat, including DoNotPay's bot negotiating with Comcast's AI.
- •Anti-bot defenses are circumvented by using each user's IP in the app
- •Big companies use AI against consumers too
- •DoNotPay's AI chats with Comcast's AI to negotiate bills
- •DoNotPay is a small drop in the bucket for companies for now
- 17:03 – 20:41
COVID, Digital Courts, and Reshaping the Legal System
Smith asks whether scaling will force companies and courts to rethink their systems. Browder says COVID accelerated legal modernization by twenty years, though old judges remain anti-technology.
- •The biggest risk to DoNotPay is that all problems get solved
- •COVID relaxed wet-signature rules, enabling products like 83(b) election filing
- •Traffic tickets are moving back into physical courts
- •Judges can be prejudiced against Zoom appearances
- 20:41 – 23:57
Robot Lawyer in the Courtroom and Legal Industry Backlash
Browder explains the speeding-ticket experiment using AI whispering through glasses, and the threats from state bar associations over unauthorized practice of law. He argues AI is not a person and says lawyers resist out of fear of losing work.
- •Offer to pay tickets if users let the AI represent them via bone-conduction glasses
- •About 300 people volunteered and two were accepted
- •State bars threatened prosecution for unauthorized practice of law
- •Browder argues AI is a technology, not a person, and compares lawyers to dinosaurs suing to stop the Ice Age
- 23:57 – 26:36
Lawyers Using AI, Mistakes, and Cross-Border Help
Browder says lawyers and judges already use AI but keep the savings, and that humans err too. He describes cross-border cases like a Mexican timeshare and how feeding the model documents improves it.
- •A Colombian judge used ChatGPT to write an opinion
- •AI mistakes are compared with human lawyer misconduct
- •AI helped cancel a Mexican timeshare under a five-day law
- •Generic models are weak at law and improve when fed many documents
- 26:36 – 27:54
Training the AI: Lawyer Review, Honesty and Guardrails
Asked about human involvement in training, Browder says top lawyers advise DoNotPay. He describes two problems with the AI: it lies and it talks too much, requiring retraining and guardrails.
- •Lawyers at firms such as Wilson Sonsini help the team
- •The AI invented facts like internet outages, creating liability
- •It talks too much, so other AIs decide whether to speak at all
- 27:54 – 30:27
What It Takes for AI Law to Be Accepted, and Deflation
Browder argues that because over 80% of people can't afford legal help, courts will have to modernize. He also discusses deflationary effects across services, since customer service costs could shrink.
- •Over 80% of people who need legal help can't afford it
- •People will have AI general counsel on their shoulder
- •Customer service may be 12% of corporate spending; AI could cut it sharply
- •Deflation could reach services unrelated to law, like airline tickets
- 30:27 – 33:46
Responding to Recent Criticism of DoNotPay
Smith raises viral criticism that the services are low quality or not truly AI. Browder says DoNotPay is moving from templates to AI, that critics created fake cases, and that the focus is consumer rights where risk is low.
- •DoNotPay is transitioning from a template company to an AI company
- •Critics signed up and created fake scenarios, like a $100,000 deposit claim
- •Goal is for the average person not to need a lawyer
- •A team member's speeding ticket case may test AI in court
- 33:46 – 38:08
AI Writing Law, Terms and Conditions, and Why Law Is Complex
Browder says AI is likely already helping write laws. He discusses DoNotPay's terms-and-conditions tool, lobbying, and his worry about anti-AI bot laws, noting existing rights like credit report disputes still apply.
- •Browder thinks AI is secretly helping draft laws via congressional staffers
- •The ChatGPT-based tool flags non-standard terms in documents like leases
- •Big-company lobbying shapes complex laws
- •He worries about laws requiring bots to disclose themselves
- 38:08 – 40:41
Supreme Court Offer and Supportive Lawyers
Smith asks about the offer of a million dollars to a lawyer with a Supreme Court case to use AirPods. Browder explains it was meant to show AI isn't scary, and says most lawyers are supportive.
- •Browder offered $1 million for a Supreme Court case with AI in an earpiece
- •He hoped it would show AI can handle documents well
- •Lower court lawyers accepted, but none from the Supreme Court
- •He talks with a San Diego judge about helping self-represented people
- 40:41 – 43:15
Future of Law: Instant Class Actions and Suing Spam Callers
Browder imagines instant class actions where robot lawyers recover money in the background. He then describes a product that uses the Telephone Consumer Protection Act to sue spam callers for $1,500 per call.
- •Class actions currently take years and pay tiny checks
- •Robot lawyers could win refunds almost immediately
- •A card trap captures spam callers' details to sue them
- •One user paid for a new roof with the proceeds
- 43:15 – 46:14
Why Horizontal Works: DoNotPay's Strategy
Asked why no one else has built this, Browder credits a horizontal strategy: single-issue products can't retain users. He says seven years of building tools for building products set DoNotPay apart.
- •a16z was the pre-seed investor, allowing him to stay at Stanford
- •People get parking tickets only about once a year, so a single-use company can't retain customers
- •Horizontal coverage built a retentive customer base
- •The team builds tools to build products so it can respond within weeks
- •Smith notes horizontal keyword research can guide the next product
- 46:14 – 49:09
Strange Laws Game and Closing
Smith quizzes Browder on true or false obscure laws, such as donkeys in bathtubs and New Jersey gas pumping. They discuss how old laws never expire and some obscure ones are still enforced.
- •Arizona donkey-in-bathtub law is true; coins-in-ears is a Hawaii law, not Rhode Island
- •Ice cream in back pocket is a Kentucky law, not Alabama
- •No process exists to expire old laws
- •New Jersey bans self-service gas, enacted in 1949
