Garry Tan, the prominent CEO of Y Combinator, has openly discussed his significant reliance on artificial intelligence agents, revealing an annual expenditure of up to $200,000 on AI tokens. This substantial investment underscores his belief in the transformative power of AI, which he integrates deeply into both his professional responsibilities and personal routines. Tan's approach highlights a vision where AI agents act as constant, intelligent companions, enhancing productivity and offering proactive assistance across various aspects of life, from managing emails and notes to aiding in complex coding projects and organizational operations. He also emphasizes the critical need for these advanced AI tools to become more affordable and remain under individual user control, steering clear of centralized corporate dominance.
Y Combinator CEO's Deep Dive into AI Agents: Costs, Capabilities, and Future Vision
In a recent conversation on Walter Isaacson's podcast, "The Builders," which aired on Tuesday, November 2, 2026, Garry Tan, the head of Y Combinator, disclosed his considerable financial commitment to artificial intelligence. He revealed that his annual spending on AI tokens ranges from $100,000 to $200,000. Tan enthusiastically detailed how these AI agents, which he jokingly suggested might cause "cyber psychosis" due to his intense engagement, are fundamental to his daily operations. He has even made his collection of 23 Claude Code skills, known as gstack, publicly available on GitHub to demonstrate the capabilities of these AI tools.
Tan's use of AI extends beyond typical work-related research, such as preparing for interviews. His AI agent manages a comprehensive database of his personal information, including all his emails, Dropbox files, Apple Notes, and even therapy records dating back two decades. He metaphorically describes this relationship as having a "Jiminy Cricket" constantly by his side, anticipating needs and offering guidance. Within Y Combinator itself, an internal AI agent named Stray Light possesses a distinct personality and an extensive database encompassing every founder the accelerator has supported, all communications, and applications, effectively streamlining operational processes.
Despite his fervent advocacy, Tan pointed out a significant hurdle to widespread AI adoption: the prohibitive cost. He acknowledged that his spending represents a considerable privilege not accessible to everyone. He expressed a desire for a future where personal AI agents are both affordable and controlled by individual users, contrasting this with the current model of large technology companies managing powerful algorithms, which he believes often leads to user discomfort regarding control and privacy.
Garry Tan's profound engagement with AI agents offers a fascinating glimpse into a future where artificial intelligence is not merely a tool but an integral, autonomous partner in daily life. His willingness to heavily invest in and openly discuss his AI usage not only highlights the immediate potential of these technologies but also sparks important conversations about accessibility, cost, and user control. As AI capabilities continue to advance, the challenges of making such powerful tools affordable and ensuring they serve individual interests rather than corporate agendas will become increasingly critical. Tan's vision of a personal, programmable AI could indeed be a cornerstone for future technological development, provided the economic and ethical considerations are thoughtfully addressed.
