Garry Tan on Startups, AI, and Keeping Markets Open

🚀 How Y Combinator chooses founders

• YC receives roughly 80,000–90,000 applications a year for about 800 places; only a small share reach interviews.
• Final interviews last 10 minutes—a deliberate constraint. Tan says founders should quickly explain:
  1. Who the user is and what problem they have.
  2. How the product solves it and why it is compelling.
  3. If there’s time, why this founder is the right person to build it.
• A clear, impressive demo can be enough: Tan says his own startup, Posterous (“dead simple blogs by email”), was understood quickly by YC’s founders and funded.
• A key test for YC is whether it could recognize and back an exceptional founder—such as Elon Musk, Bill Gates, Steve Jobs, or successful YC alumni—in a short interview.

🧭 Merit, access, and Tan’s experience

• Tan says YC is not a perfect meritocracy, but should keep striving toward one.
• He argues that funding can give people a chance to build without relying on wealthy families. YC’s investment has grown from $18,000 when he participated to $500,000 today.
• Before becoming an investor, Tan worked at Palantir and founded Posterous. He says firsthand startup experience helps investors understand what founders face.
• At Palantir, Alex Karp taught him that presentation matters: the company put young engineers up at the Mandarin Oriental because the quality of their surroundings should match the quality of the software they were presenting to senior customers.

⚖️ Antitrust, competition, and YC’s policy role

• Tan argues that self-preferencing and closed platforms can block competition, and that open markets are essential for startups.
• He points to the history of Microsoft’s dominance in browsers—and government action against it—as part of the context in which Google later grew.
• He supports measures such as Europe’s Digital Markets Act and open-app-market policies, including greater portability, transparency, and interoperability.
• He sees “little tech” and startups as a counterweight to concentrated power in big tech, and says AI companies could become a new generation of powerful gatekeepers.
• YC has begun engaging more actively in policy, Tan says, because tech policy is shaped even when startup founders are absent. He describes YC’s advocacy operation as small but intended to give startups a collective voice.

🌍 Immigration and personal background

• Tan says about half of major successful startups have at least one immigrant co-founder, and sees immigration as a distinctive source of American entrepreneurship.
• His mother’s family came from Burma as refugees; his father came from Singapore.
• He also describes growing up with uncertainty around his father’s unemployment and alcoholism. He believes the experience made him highly attentive to other people’s emotions—a useful skill when working with founders.

🤖 AI agents: more than chat

• Tan sees AI agents as a potential core operating system for work, not just a feature or chatbot.
• YC’s internal agent, Straylight, has access to extensive company context, including founder information, applications, and correspondence.
• He says agents can hold and synthesize far more information than a person can keep in working memory, and may increasingly act proactively—anticipating needs and helping users “see around corners.”
• As an example, he says his agent reviewed his email, Dropbox, notes, and past therapy notes, and helped prepare for the interview by transcribing and analyzing the interviewer’s past podcasts.
• Tan acknowledges that this kind of agent use is currently expensive; he estimates his own AI-token spending at $100,000–$200,000 a year.

💼 AI, jobs, and responsibility

• Tan hopes AI will ultimately create new work, but acknowledges that some jobs will be displaced and that retraining will be necessary.
• He promotes teaching people to use AI tools—citing his open-source G Stack project, designed to help developers produce much more software with tools such as Claude Code.
• His proposed path is to combine domain expertise with AI capabilities: for example, someone experienced in medical billing could work with a technically skilled person to build a company in that field.
• He urges people to identify what they uniquely know from their own lives and use AI to act on that knowledge.
• Tan argues that leaders should use AI to expand what people can do—not simply to cut costs—and says ethical judgment matters when choosing which companies to fund. He cites a startup involving ADHD medication for children as an example of a potentially harmful business model that YC did not back.

🔓 Personal AI, open source, and the future

• Tan imagines personal AI agents giving users more control over what they see online, instead of leaving recommendation algorithms entirely in the hands of platforms.
• In his vision, people could use an agent as a personal filter for social media—shaping feeds around their own goals and, potentially, input from people close to them.
• He predicts a coming competition over which AI agent people use, similar to an operating-system or browser war. He worries that a few companies could control a large share of future commercial activity.
• He wants an easy, affordable open-source “Linux for personal AI” so people can build alternatives to concentrated AI platforms.
• Tan supports regulation that keeps markets open, while rejecting both a big-tech oligarchy and policies he views as excessive state seizure. He calls for a middle path that enables competition, new businesses, and broad prosperity.

🔑 Central message

Tan’s overarching argument is that innovation depends on giving founders access to capital, talent, and open markets. He sees YC as a way to find promising builders—and increasingly as a voice for them in policy debates—while arguing that AI’s benefits will depend on keeping the technology accessible and using it responsibly.