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 a fundamental shift from static documentation to iterative, test-driven development. Evals have replaced traditional PRDs as the primary mechanism for defining user value, allowing teams to quantify model performance and address specific pain points like hallucination or instruction following. Success in this environment demands that product leaders remain hands-on, "sweating the tokens" alongside engineers to uncover emergent capabilities and refine product-market fit. Anthropic’s Labs organization exemplifies this approach by incubating discontinuous bets—such as Claude Code and computer use—through a bottom-up culture that prioritizes experimentation over rigid planning. While AI excels at execution, human judgment remains critical for navigating ambiguity, setting strategic direction, and ensuring that models remain aligned with user needs and safety standards.