A major earthquake has rocked the AI world: OpenAI has snatched a key product from its arch-rival, Anthropic. Cat Wu, the former head of product at Claude Code, who just announced her departure yesterday, has today officially joined OpenAI to lead a core product team. What does this mean? It's like the captain of Barcelona joining their arch-rival, Real Madrid. OpenAI President Greg Brockman quickly retweeted a welcome message three minutes later, and OpenAI executives lined up in the comments section. Isn't Anthropic about to go public? Why are these core product leaders leaving so quickly? From October 26th, when Anthropic secretly filed its IPO prospectus, in just 12 days, five key figures have left the two flagship teams, Claude Code and Cowork. One went to Musk's xAI, the head of product joined their arch-rival, OpenAI, and the rest went solo. Because top AI product managers have realized: building applications on someone else's model leaves you constrained. To define the next generation of agents, we must return to the labs with their own top-tier, self-developed models. But the real signal behind this is extremely jarring: In the AI battlefield of 2026, the model foundation is no longer the moat; the product experience is. Both companies have equally intelligent foundations, but whoever has a more user-friendly product, whoever allows white-collar workers to command AI, will win customers. Silicon Valley's current talent war is no longer about acquiring mathematicians. It's about acquiring people who truly understand users and can turn models into convenient tools. Models will eventually become incredibly cheap; the real value lies in the person who understands how to use them. Therefore, Deepseek is also currently developing the Harness app. Do you see the investment opportunity here?