Inside OpenAI: 2026 is the year of agents, AI’s biggest bottleneck, and why compute isn’t the issue

In this episode of Lenny's Podcast, Lenny Rachitsky interviews Alexander Embiricos, product lead for Codex at OpenAI, about the development and future of Codex, OpenAI's coding agent. Embiricos discusses Codex's explosive growth, its architecture, and its role as a software engineering teammate, emphasizing its potential to accelerate coding and other tasks. He also touches on the impact of AI on engineering jobs, the importance of user feedback, and OpenAI's approach to building AGI, highlighting the significance of human typing speed and review as limiting factors. The conversation explores real-world applications of Codex, such as the development of the Sora Android app, and delves into the vision of Codex as a super assistant capable of coding and performing various tasks to enhance productivity.