Garry Tan's AI agent spending

Y Combinator CEO Garry Tan spends about one hundred thousand to two hundred thousand dollars a year on tokens for AI agents, he told Walter Isaacson's podcast The Builders.

“I spend about maybe $100,000 to $200,000 a year on tokens”

How Tan uses his agents

Tan uses agents for work projects, coding tasks, and personal life management. His agent reads all his email, Dropbox files, Apple Notes, and therapy notes going back twenty years.

“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”

Agents as personal assistants

Tan describes AI agents as autonomous helpers that know you deeply and help with daily tasks. He compares them to a Jiminy Cricket on your shoulder, helping you see around corners.

“Think of it as a Jiminy Cricket on your shoulder at all times”

Treating agents differently from software

Tan warns against treating AI agents like traditional software. Focusing too much on repeatability and safety trades off the benefits of working with systems that behave like humans.

“they actually trade off all of the beautiful parts of working with an AI agent, which basically can be so similar to a human”

Y Combinator's internal agent

Y Combinator uses an internal AI agent named Stray Light that has personality and company knowledge. It holds context on every founder backed, all emails received, and all applications.

Tan's shared agent skills

Tan has shared his set of twenty three Claude Code skills called gstack on GitHub to teach others what agents can do. He is an early OpenAI adopter and power user.

The cost barrier to mainstream adoption

Tan says AI agent costs are too expensive for mainstream adoption. He acknowledges his spending is a privilege not everyone can afford and costs must fall significantly.

“It's too expensive right now”

Personal AI over Big Tech control

Tan advocates for personal AI agents users control and program themselves rather than free agents controlled by large technology companies like Meta that control algorithms.

“a personal AI that you control, that you program”