Asana has shared a striking example of AI-powered software development: the company's engineering team used OpenAI's Codex agent to complete a frontend test migration that was projected to take at least five years — finishing it in approximately two weeks.
The migration involved moving Asana's entire frontend testing infrastructure to a modern framework. According to the company, the work would have cost an estimated $6 million if done manually, but was completed for roughly $12,000 using Codex.
The announcement, shared by Asana on Instagram and subsequently covered by tech outlets, has sparked debate about the pace at which AI coding tools are compressing traditional software engineering timelines. Some engineers have expressed skepticism about the full scope of what was migrated, while others point to it as a concrete signal of how AI agents are beginning to automate substantial portions of development work.
OpenAI Codex is an AI coding agent that can autonomously write, test, and iterate on code. It operates within a sandboxed environment and can handle multi-file engineering tasks with minimal human oversight.
The story adds to a growing body of evidence that AI coding tools are no longer limited to simple boilerplate generation and are increasingly tackling complex, large-scale engineering projects.
