At Anthropic, the primary artifact for product managers is no longer the PRD. Instead, they create "evals" (evaluation sets) from user feedback to define problems and measure model improvements. This makes user needs directly actionable for AI researchers, changing a core PM workflow.
Anthropic realized that a powerful model like Opus 4.5 only achieved mass adoption when paired with an innovative product like Claude Code. The magic of a frontier AI is lost if users cannot easily access its advanced capabilities through a well-designed, intuitive product.
In the fast-moving AI space, product leaders cannot just manage. Dianne Penn insists that senior PMs and she herself must stay hands-on by owning workstreams and shipping. This is crucial for developing taste and understanding the technology's evolving capabilities to effectively guide their teams.
Contrary to the idea that safety features make an AI boring, Anthropic's "constitution" makes Claude more useful. It's designed to push back and challenge ideas, acting as a true thinking partner rather than a sycophantic assistant. This leads to better outcomes and a more interesting user experience.
While heavy token spend can help you "live in the future," real value comes from communal discovery. Anthropic found that having the entire company experiment in a public Slack channel led to magical, emergent use cases as people rapidly built upon each other's ideas.
AI models experience sudden, discontinuous jumps in specific capabilities—a "jagged edge." The product role must shift from following a predictable roadmap to actively discovering these new, often unexpected, abilities and rapidly building product experiences around them.
To avoid building products that will quickly become obsolete, Anthropic's PMs design for future model capabilities. By asking, "What will Claude 8 enable for users?", they make architectural and UX decisions today that will be compatible with AI's exponential progress.
Dianne Penn uses a custom Claude "skill" based on the book "Crucial Conversations" to prepare for difficult management talks. The AI acts as a personalized coach to brainstorm approaches and anticipate reactions, augmenting emotional intelligence (EQ) in addition to intellectual tasks.
To prevent over-reliance on AI and maintain independent thought, form your own point of view on a topic first. Then, use the AI as a sparring partner to augment or challenge your thinking, rather than letting it do all the initial work and dictate your perspective.
The most successful researchers at Anthropic, including senior leadership, remain deeply embedded in the details. They personally review training run progress, analyze evals, and examine underlying data. This hands-on approach is critical for developing taste and making breakthroughs in a rapidly evolving field.
To avoid burnout amid rapid model releases (shipping more in Q2 than all of 2024), Anthropic fosters a culture where team members have radical ownership but also actively support each other. This "hive mind" approach allows individuals to take time off without work piling up, sustaining momentum.