Dropping out of Yale for a startup: Dawson Chen's journey to Martin AI

Executive overview

Most students wait years before attempting a startup. Dawson Chen got into Y Combinator during his freshman year at Yale and left the next day.

He's building Martin, a personal AI butler designed to know you deeply — handling email, calendar, and voice interactions like a long-tenured human assistant.

Young founders accumulate startup experience faster than peers who take the corporate route — the perceived disadvantage is largely illusory.

The decision to drop out

• Got into YC during freshman year; left Yale the day after
• Had spent all of second semester looking for a reason to drop out
• Saw YC as "a ticket to skip the entire line" to starting a company
• Felt three more years of school wouldn't accelerate his startup goals
• By 25, a founder-track 19-year-old will have more relevant experience than someone who spent those years at a big company

Early projects and building instincts

• Grew up in Mountain View; both parents were software engineers
• Started coding obsession in fourth grade; took every CS class available in high school
• First startup: a social app for song parodies — took three months to build, never launched properly, no users
• At Yale: built a campus tours marketplace (high schoolers pay Yale/Harvard/Princeton/MIT students for personalised tours)
• Recruited customers by camping outside the admissions office — reached four-digit MRR
• The tours business wasn't scalable, but it proved he could find and convert real customers

Getting into Y Combinator

• Applied two months past the deadline on a whim ("it takes an hour, I'll just apply")
• Applied with a different idea: AI customer support chatbots for grocery stores
• Got two interviews with Jared; heard back quickly; attended retreat about a week after applying
• Felt immediate community fit on the first day — "member community fit" in his own words

What YC taught him

• Ruthless focus: wake up, work all day on one thing (improving product for first 100 users), sleep, repeat — seven days a week
• At the early stage, everyone is struggling; there's no instant viral hit
• It's normal to not have a solid plan — launch things and listen to users
• Don't be afraid to not have all the answers
• Pivoted inside the batch; spent a couple of weeks exploring ideas before landing on Martin

Why Martin and what it is

• Inspired by Jarvis from Iron Man — wanted it since watching the first film as a kid
• Identified the gap: current AI assistants have privacy concerns and don't truly learn you
• A great assistant knows 95% of what you want from 5% of the command, built on years of context
• First version: one button, voice input, simple memory, personality, note export
• Launched September to 2,000 users; iterated intensively with them
• March launch adds email integration, calendar, and phone interaction

Learning from the first user — his dad

• Father used Martin on a 15-minute walk after dinner every night
• Gave unfiltered feedback: "got kind of bored" or "he said something great today"
• Revealed that personality and character of the agent matter as much as utility
• Also surfaced the importance of memory and proactive behaviour — knowing a user's worries, asking the right questions
• Key shift: move from founder's vision to solving the stranger's problem

Building and hiring philosophy

• Lives in a hacker house in SF; works 9am to 11pm–midnight daily
• Keeping the team extremely lean to stay flexible as the product evolves
• Looking for early people brave enough to experiment, not those who need certainty
• Most early growth from word of mouth; deliberately avoided heavy marketing until retention was solid

Advice for young founders

• You are not as disadvantaged as you think — startup experience compounds fast
• Don't be afraid to disappoint people or ignore advice from those who haven't done startups
• The industry is counter-intuitive; standard advice often doesn't apply
• Build confidence by facing hard challenges directly, not by waiting until you feel ready