YouTube02 Dec 2025
42m

Anthropic's First PM: Opus 4.5, Rethinking Model Scaffolding & Safety as a Competitive Advantage

Podcast cover

Unsupervised Learning: Redpoint's AI Podcast

Anthropic's Opus 4.5 model is the focus, with Dianne Na Penn, head of product for research at Anthropic, discussing its capabilities and the company's research process. Anthropic aims to improve instruction following, coding, and memory in its models. The company focuses on practical applications, such as enhancing Claude's abilities with Excel and PowerPoint, which resonated with financial services customers. Computer use has evolved from an experimental feature to an end-to-end agent, capable of tasks like calendar rescheduling. Opus 4.5 has proven helpful for product teams in pricing and positioning, offering efficiency gains and intelligence at a lower cost. Anthropic prioritizes business use cases, focusing on data security and enterprise requirements, and emphasizes the benefits of aligned, safe models for higher-value intelligence.

Outlines

Part 1: Roadmap, Vision, and Opus 4.5

Part 2: Capabilities, Pricing, and Market Education

Part 3: Agents and Enterprise Strategy

Part 4: Long-Running Intelligence and Evaluations

Part 5: Scaffolding and Model Autonomy

Part 6: Culture and Internal Journey

Part 7: Future Outlook and Model Taste

Part 8: Safety, Alignment, and Conclusion

Sign in to continue reading, translating and more.

Open full episode in Podwise