Google Cloud poaches Harness engineering leader from OpenAI, bets on Enterprise Agent Workflow

According to Dynamic Beating monitoring, former OpenAI engineer Ryan Lopopolo has announced joining Google Cloud as the Agentic Google Cloud Platform Chief Engineer.

Lopopolo was the author of the OpenAI article "Harness Engineering." This methodology emphasizes configuring Agents with context, tools, permissions, evaluation, and feedback loops to actively engage in software development.

He participated in an internal experiment at OpenAI where the team used Codex to build a product from an empty repository, generating approximately 1 million lines of code in 5 months. The code, testing, CI, documentation, and internal tools were all completed by Codex, with humans mainly focusing on goals and acceptance criteria.

This job move indicates that Google Cloud aims not only to sell models and compute power but also to enhance enterprise Agent engineering capabilities.