Y Combinator's Garry Tan Runs His Life on AI Agents — for Up to $200,000 a Year in Tokens
The startup accelerator chief told a podcast he spends as much as $200,000 a year on AI tokens, using agents to run his work, his inbox, and even his decades of personal notes.

Y Combinator chief executive Garry Tan has put a price on his enthusiasm for AI agents: somewhere between $100,000 and $200,000 a year in tokens alone, according to TechCrunch's account of his appearance on Walter Isaacson's podcast, The Builders.
An agent that knows everything
Tan has previously joked about suffering from "cyber psychosis" and barely sleeping because he is so excited to work with his AI agents. On the podcast, he described an agent that has absorbed nearly his entire digital life.
"It’s read all of my email and all of my Dropbox. It’s read, like, all of my Apple Notes, even my therapy notes going back 20 years," he told Isaacson, according to TechCrunch.
He uses agents the way many people use AI assistants — researching podcast guests and preparing for questions he might be asked — but also as a constant companion for daily life. He likened the experience to having "a Jiminy Cricket on your shoulder at all times," an agent that knows its user, runs ahead with suggestions, and helps its user see around corners. He has also published his set of 23 Claude Code skills, called gstack, on GitHub to show others what agents can do.
Inside Y Combinator, an agent with a company brain
The experiment extends to the accelerator itself. Tan said Y Combinator is using an internal agent named Stray Light that is rebuilding how the organization operates. The agent has a personality and what he called a company brain — the total context of every founder the accelerator has ever backed, everyone who has emailed it, and all of its applications, according to the report.
As an early adopter and now power user of the agent framework OpenClaw, Tan argued that people make a mistake when they treat agents like conventional software. Worrying too much about repeatability and safety, he said, trades away the qualities that make agents feel almost human to work with.
The catch: cost
For all his enthusiasm, Tan's main prediction about personal agents comes with a caveat. They will eventually go mainstream, he believes, but only if the price collapses.
"It's too expensive right now," he said, adding that he recognizes the amount he spends is "a crazy privilege that not everyone can do," according to TechCrunch.
He is also wary of the alternative in which free agents are controlled by giant platforms. Reflecting on Meta's influence as a social media company, he said the future he wants is a personal AI that the individual controls and programs — not one steered by a company that controls the algorithm.
For now, Tan's setup remains an outlier: a six-figure token habit, a decades-deep personal archive handed to an agent, and an accelerator increasingly run with an AI colleague. Whether that becomes normal, he suggests, depends less on the technology than on the bill.
Written by the BidAsk Newsroom. Based on reporting from the sources below.
Sources
Which AI tool is actually worth your money?
ToolROI is a free AI-tool ROI calculator for freelancers — compare what tools like Jasper, Copy.ai, Kit, and Semrush really cost you versus what they earn back.
Calculate my ROI →Free tool run by the same publisher as BidAsk. No affiliate links.
