For Y Combinator CEO Garry Tan, the enthusiasm for artificial intelligence goes far beyond casual adoption. Tan’s deep dive into autonomous software has reached a scale where his personal operational costs rival small engineering budgets.
Speaking on an episode of Walter Isaacson’s podcast, The Builders, Tan pulled back the curtain on how heavily he relies on AI agents to run his daily workflows and handle complex tasks.
Heavy Personal Investment in Autonomous Agents
That high expenditure reflects an intensive, around-the-clock engagement with automated workflows. Tan has previously joked that his intense excitement for working alongside AI agents has occasionally led to sleepless nights and what he playfully describes as "cyber psychosis."
The Shift Toward Agentic Workflows
Tan’s massive token consumption highlights a broader shift happening among technology executives and startup founders within the Y Combinator ecosystem. Rather than merely using basic chatbots for simple prompts, tech leaders are increasingly turning to multi-step AI agents capable of reasoning, executing code, and managing multi-faceted projects autonomously.
While spending six figures annually on tokens is far outside the norm for typical software users, it underscores how deeply power users are stress-testing the limits of current large language models to maximize personal and organizational productivity.
As AI tools continue to evolve, high-volume users like Tan provide an early look at what hyper-automated executive workflows might require in terms of infrastructure, cost, and computing power.
