OpenAI: "AI Colleagues" and Two-to-Three Month Planning

Tara Seshan, product lead for Codex and ChatGPT Work at OpenAI, describes a phase where persistent AI agents ensure execution, shifting the advantage towards human ambition and judgment. She also mentions a planning horizon for model capabilities of two to three months and internal codes focused on usage.

Ambition as a Bottleneck and PM Responsibility

Tara Seshan presents ambition as a new bottleneck for companies, in a context described as a third era of AI marked by persistent AI colleagues. According to her, elevating the ambition of others becomes a key part of the product manager's role. She also addresses writing as an act of reflection rather than just a report, in the way of leading this work. This framework accompanies the shift of value towards human judgment and the elevation of collective goals.

Two to Three Month Horizon for Model Capabilities

OpenAI plans to develop model capabilities with a two to three month lead time. Internal expressions such as “Is this maximized?” and “Are you already consuming it?” illustrate a culture focused on the concrete use of advancements. This change is accompanied by a shift from “rowing” to “steering,” where human judgment and ambition become differentiating factors as AI takes over execution.

Current Roles and Tara Seshan's Journey at OpenAI

Tara Seshan serves as the product lead for Codex and ChatGPT Work at OpenAI, while Andrew Ambrosino holds the technical lead. She spent over six years at Stripe, where she was among the first five product managers. She then led product at Watershed, which was named one of Time's Best Inventions of 2022. She is also a founder, a Thiel Fellow, and one of the first three Lenny’s Newsletter Fellows. This journey sheds light on the perspective she develops on organizing work with AI.