Summary
Akshay Nathan, who leads core product engineering at OpenAI, sits down with the Latent Space podcast to discuss the vision behind ChatGPT Work and OpenAI’s broader push to make ChatGPT an everything app — a single surface where knowledge work, code execution, research automation, and collaborative data analysis converge.
Nathan traces his own career arc from consumer fintech through no-code tools at Airtable to OpenAI, framing LLMs as the missing piece that finally makes the democratization-of-code thesis achievable. He is candid about what has and hasn’t changed since he joined OpenAI in 2023 when the company had 500 people: the startup energy and bottoms-up ambition have remained constant even as the product surface has expanded dramatically across ChatGPT, Codex, and enterprise offerings.
The most concrete segments involve live demos of ChatGPT Work in action — using the platform to train competing AI agents to play a game against each other, run auto-research loops, generate benchmarks, and publish findings as interactive sites rather than markdown documents. Nathan argues that as building gets cheaper and faster, the real bottleneck becomes ideas and taste rather than engineering capacity, which has implications for how product teams should think about generalists, user feedback loops, and what kinds of roles remain uniquely valuable in an AI-saturated environment. The episode offers rare visibility into how OpenAI’s product team thinks about the platform’s long-term trajectory.
📺 Source: Latent Space · Published July 28, 2026
🏷️ Format: Podcast
