Bret Taylor of Sierra on AI agents, outcome-based pricing, and the OpenAI board

Bret Taylor has been inside more platform shifts than almost anyone in Silicon Valley—Google Maps, the Like button, Salesforce, the Twitter acquisition, and now the OpenAI board. His day job is CEO of Sierra, an AI customer experience company that hit $165 million in ARR in just over two years. In this conversation, he explains why the messiest AI agents might be the most useful, why outcome-based pricing is the right model for applied AI, and why the hardest part of building an AI company is knowing when to throw your own work away.

Why coding agents work (and why most other agents don't—yet)

Taylor points out that coding agents have undergone a transformation in the past four months that is rare for any technology. But he cautions against extrapolating too quickly to other domains. Code repos have specific qualities that make them ideal for AI: all context lives in one place, in text files, with built-in feedback loops like compiler errors, unit tests, and code reviews. "It's almost designed for a robot," he says.