The conversation centers on the evolving landscape of agentic engineering and the shifting strategies of major AI labs. OpenAI and Anthropic face scrutiny for their opaque communication and compute constraints, while the emergence of Thinking Machines and their open-weight model, Inkling, signals a potential pivot toward enterprise-customized, proprietary models. A significant portion of the dialogue focuses on practical, terminal-based AI workflows, specifically the use of minimal agent harnesses like PAI, the Herder multiplexer, and the Planitator tool for local code review and annotation. These tools allow developers to maintain control over system design and type-driven development, countering the tendency of automated agents to produce suboptimal abstractions. The speakers also address the industry-wide challenge of developer burnout, reflecting on the exhaustion inherent in navigating rapidly changing AI tooling and the need for intentional, sustainable development practices.
Sign in to continue reading, translating and more.
Open full episode in Podwise