OpenAI Leader Questions Need for Product Managers
On the 20VC podcast, OpenAI's Alexander Embiricos suggested that the Product Manager role is not always strictly necessary, especially on smaller, high-context teams. He argued that strong engineering leads or product-minded designers can often fulfill PM functions, as the role is explicitly undefined and adapts to team needs. This challenges traditional organizational structures for product development in the AI era.
- OpenAI's engineering teams are often structured as small, autonomous "pods" of about six people that operate like mini-startups. A typical pod is led by a technical "Pod Owner" rather than a traditional manager and includes senior engineers, machine learning specialists, and a quality assurance expert. - Alexander Embiricos leads the product team for Codex, the AI model that powers tools like GitHub Copilot. He joined OpenAI after it acquired his developer collaboration startup, and his perspective is shaped by building AI tools for highly technical users. - Embiricos has predicted that within the next two years, the majority of code will be written by autonomous AI agents, shifting the developer's role to that of an orchestrator. This shift underlies the argument for rethinking traditional software development team structures. - The concept of a "product-minded engineer" involves developers who are deeply interested in the business and user context behind what they are building. They proactively contribute product ideas, challenge specifications, and suggest alternative approaches to better balance user impact and engineering effort. - This view is not universal within the company; OpenAI product lead Britt Jamison has stated that as AI takes over more routine tasks, product managers will be promoted into more strategic roles rather than being eliminated. - The broader shift in AI product development is from managing "deterministic" products with predictable, linear user flows to "stochastic" or behavioral products. In this new model, a PM's role becomes defining behavioral goals and guardrails for an adaptive AI rather than specifying exact outputs. - OpenAI's product strategy has historically prioritized speed and rapid iteration over polish, as seen when it shipped the initial version of ChatGPT in weeks to refine it with real-time user feedback. - The company operates under a unique structure where a non-profit foundation controls a for-profit public benefit corporation (PBC), a model intended to ensure its mission of benefiting humanity guides its commercial product strategy.
