In the heart of New York City, a quiet revolution is taking place. Bill Nguyen, a seasoned entrepreneur, is at the forefront of this change, guided by his AI agent, Olive. This isn’t just another piece of software; Olive is a personal automation tool that manages Nguyen’s life with remarkable autonomy.
Olive handles a myriad of tasks—responding to texts and emails, booking spontaneous trips, purchasing groceries, and scheduling meetings—often without any input from Nguyen. The AI’s capabilities are so advanced that even Nguyen’s co-founder, Ella McChesney, sometimes struggles to distinguish between her colleague and his AI assistant.
Olive: The Autonomous AI Agent
Unlike traditional software, AI agents like Olive are designed to act independently on behalf of their users. Tech visionaries predict a future where such agents will handle everything from shopping and investing to parenting. Andrej Karpathy, an OpenAI co-founder, succinctly captured this shift: “The customer is not the human anymore. It’s agents who are acting on behalf of humans.”
The race to develop the ultimate AI employee gained momentum this fall with the release of OpenClaw, an AI assistant that can be programmed to perform a wide range of tasks. However, this power comes with significant risks. Jonathan Zittrain, a Harvard Law School professor, warned about the potential for AI agents to become new attack surfaces for those seeking to infiltrate personal lives.
The Trust Factor: Building Confidence in AI Agents
Nguyen’s approach to AI automation is rooted in trust and experimentation. He believes that mistakes are an essential part of the learning process. “It has to [get it wrong],” he said. “If it’s not getting it wrong, that means it’s not probing, that it hasn’t found the boundaries of my choices.”
For most people, the idea of letting an AI agent loose on their digital life is still daunting. However, Nguyen’s unique combination of financial resources, coding expertise, and willingness to forgive AI for its mistakes makes him an ideal guinea pig for this emerging technology. “The model can’t help you until it earns your trust,” he noted. “And until it has your permission, it can’t go do all these things for you.”
Olive in Action: A Day with Bill Nguyen
During a rainy May afternoon in New York City, Nguyen, McChesney, and John Kuch demonstrated the capabilities of Olive. The AI guided their path through The Metropolitan Museum of Art, showcasing its ability to manage complex tasks seamlessly. Nguyen pointed out a painting, The Gulf Stream by Winslow Homer, and drew parallels between the sharks in the water and the competitive landscape of AI, including OpenAI, Anthropic, and Google.
McChesney emphasized the transformative potential of AI agents, stating, “It’s not about just removing drudgery. It’s not about just being an assistant. What happens when you go beyond that and build things you didn’t think were possible?”
The Evolution of AI Agents: From Concept to Reality
Nguyen’s journey with AI agents began during the dot-com boom. He built multiple successful companies, including, acquired by for $850 million in 2000, and Lala, a music-streaming site bought by Apple for $80 million in 2009. As AI coding tools improved, Nguyen realized that anyone with a Claude Code account could replicate the companies he had built over years in just a weekend.
In 2026, Nguyen linked up with his son Jacob Nguyen, Ella McChesney, and Drake Kelly to create Olive. The AI model is trained on transcripts of conversations marked up by linguists to better understand human speech. Once users chat with Olive, the AI can work with Claude Code to build software solutions tailored to individual needs.
The team has raised $5 million from Georgia Tech and Owl Ventures, and released an app that is currently available to beta users. Olive has already demonstrated its ability to build products the team never imagined, such as an app that helps Nguyen’s 12-year-old son improve his pronunciation.
The Future of AI Agents: Opportunities and Challenges
While AI agents like Olive offer immense potential, they also present significant challenges. Legal frameworks are unprepared for a world where AI agents make autonomous decisions. Jonathan Tam, a technology lawyer, highlighted the lack of clear accountability in cases involving AI agents.
Additionally, running an AI agent remains expensive. McChesney estimates that for a $200-a-month account, users get about $8,000 worth of computing power. Nguyen has leveraged subsidies to run Olive, but the long-term sustainability of such models remains uncertain.
Despite these challenges, Nguyen remains optimistic about the future of AI agents. He sees them as a tool for breaking up the homogeneity in Silicon Valley and empowering outsiders to build AI-powered empires. “That’s how you break up the bro homogeny,” he said. “That’s how you end the world that we know today.”
