OpenAI Hires Anthropic's Cat Wu to Lead Core Product Management

OpenAI president and co-founder Greg Brockman welcomed her, as did Thibault Sottiaux, who is responsible for core products at the company.

Claude Code experience

Wu spent two years at Anthropic and helped develop Claude Code, which began as a limited internal project and became an important tool for AI-assisted programming. She is known for using Claude Code to build a reinforcement learning environment, guiding the system for dozens of hours without writing the code by hand. When she left Anthropic, she thanked her colleagues and the users whose feedback shaped products including Claude Code and Cowork.

Agents that keep working

Her arrival coincides with OpenAI's push to expand what AI agents can do. The goal is for them to keep working on tasks for long periods, even after a user closes their computer, rather than only carrying out direct commands. In June 2026, OpenAI announced plans to acquire the startup Ona, which specializes in secure cloud execution, to integrate its technology into Codex. Codex has grown from software development into research, analysis and task automation.

Wu said her aim is to build agents that help individuals and teams while users keep control: directing tasks, steering them and tracking results. That is a growing challenge for the industry. Companies need to define what data agents can access, which systems they operate through, and how their steps are logged and reviewed.

By combining Wu's product experience with its cloud execution work, OpenAI is seeking to widen agent capabilities while keeping humans in charge of guidance, oversight and decisions.