Why humans are AI’s biggest bottleneck (and what’s coming in 2026) | Alexander Embiricos (OpenAI Codex Product Lead)

In this interview, Lenny Rachitsky speaks with Alexander Embiricos, product lead for Codex at OpenAI, about the development and future of Codex, OpenAI's coding agent. Embiricos describes Codex as an AI teammate for software engineers, emphasizing its potential for proactivity and integration into the software development lifecycle. They discuss the explosive growth of Codex, the importance of building a coding agent that can use a computer effectively, and the shift towards making code review easier. Embiricos also touches on the impact of Codex on engineering jobs, the potential for non-engineers to use it, and the limiting factors for achieving AGI, highlighting the need to improve human-AI interaction and validation processes. The conversation also explores the vision for Codex as a super assistant, its applications in projects like the Sora Android app and Atlas browser, and the importance of understanding customer needs in the AI era.