Anthropic’s first technical PM on token maxing, the jagged edge, and living in the future | Dianne Penn

Product management in the AI era requires shifting from static documentation to dynamic, test-driven development where "evals" serve as the new PRDs. Dianne Penn, Head of Product for Anthropic’s research and labs, emphasizes that building frontier models necessitates hands-on experimentation and "sweating the tokens" to uncover emerging capabilities. The Labs team operates by identifying discontinuous bets—such as Claude Code and computer use—that push the boundaries of what AI can achieve. Success in this volatile field demands a first-principles approach, where product leaders act as active participants in the training loop, using models as sparring partners to refine strategy. Ultimately, human judgment remains the critical differentiator, as the most effective AI products are those that augment human thinking, provide proactive feedback, and maintain a clear, cohesive vision amidst rapid technological acceleration.