Ex-OpenAI insider lifts lid on gruelling workloads and Twitter-fuelled culture at ChatGPT parent

/2 min read

ADVERTISEMENT

In a nearly 4,000-word blog post, Owen, who led the launch of Codex—an AI model that converts natural language into code—reflected on his time at ChatGPT’s parent company.
Ex-OpenAI insider lifts lid on gruelling workloads and Twitter-fuelled culture at ChatGPT parent

From fluid team structures to intense workloads, former OpenAI employee Calvin French-Owen offered a rare glimpse into the company’s internal workings.

In a nearly 4,000-word blog post, Owen, who led the launch of Codex—an AI model that converts natural language into code—reflected on his time at ChatGPT’s parent company.

He revealed that the push for Codex began in February 2025, following a team merger and a strong sense of urgency, with a public launch in just seven weeks.

Owen shed light on the intense workload, noting that one engineer even returned early from paternity leave to rejoin the effort. With late nights, early mornings, and weekend shifts, the pace was relentless. He added that the company closely monitors public opinion circulating on X (formerly Twitter).

“If you tweet something related to OpenAI that goes viral, chances are good someone will read it and consider it. A friend of mine joked, ‘This company runs on Twitter vibes.’ As a consumer company, perhaps that’s not so wrong. There’s certainly still a lot of analytics around usage, user growth, and retention—but the vibes are equally important,” said Owen.

Fortune India Latest Edition is Out Now!

Read Now

Despite the pace, the scope of Codex was ambitious. The team built a container runtime, improved repo downloading, fine-tuned a custom model for code edits, handled complex Git operations, and enabled internet access—all while creating an intuitive new user interface embedded in ChatGPT. The launch was a sleepless, high-stakes operation: a final 4 a.m. deployment followed by an 8 a.m. livestream. But it paid off.

The final product embraced an asynchronous model where users assign coding tasks and Codex returns with pull requests (PRs), like a human collaborator. Codex saw immediate adoption, delivering 630,000 public PRs in just 53 days—over 78,000 PRs per engineer.

Owen noted that the complexity and scale of innovation at OpenAI are exactly what make it a “secretive place.”

“I couldn’t tell anyone what I was working on in detail. There’s a handful of Slack workspaces with various permissions. Revenue and burn numbers are more closely guarded,” he added.

He said that seamless collaboration between teams helps improve cross-functional effectiveness across the company.

“OpenAI is perhaps the most frighteningly ambitious org I’ve ever seen. Teams at OpenAI are much more fluid than they might be elsewhere. There was no ‘waiting for quarterly planning’ or ‘re-shuffling headcount.’ It moved really quickly,” Owen added.

Fortune India is now on WhatsApp! Get the latest updates from the world of business and economy delivered straight to your phone. Subscribe now.

Related Tags