Summary
Tara Seshan, OpenAI’s product lead for Codex and ChatGPT, joins Lenny Rachitsky to discuss the evolving shape of AI products and what it’s actually like to build at a frontier lab. Seshan frames AI product development in three eras: the first centered on chat, the second on agentic task completion, and a coming third era defined by persistent AI coworkers that collaborate alongside humans over time — a shift she argues will require fundamentally different product thinking.
Seshan explains why OpenAI’s internal culture surprises new hires: rather than a centralized strategic playbook, teams operate with founder-level autonomy and thin distance from the market. She argues that the only viable planning horizon in the current environment is two to three months, and that being prolific and empirical — shipping things users can test fast — matters far more than elaborate upfront reasoning.
On the product side, Seshan unpacks the relationship between ChatGPT’s work mode and Codex, clarifying that the separation is largely a UI and onboarding decision rather than a capability split. With ChatGPT at over a billion monthly active users and Codex growing rapidly among developers, the long-term vision is a unified experience where users don’t have to choose between modes. The episode is an unusually candid look at product strategy inside one of the most closely watched companies in AI.
📺 Source: Lenny’s Podcast · Published August 30, 2026
🏷️ Format: Podcast
