OpenAI CEO Sam Altman is setting a breakneck pace with his company, rolling out feature after feature to keep the multibillion-dollar gravy train steaming ahead.

And that kind of drum beat, especially paired with ChatGPT’s meteoric rise, is bound to cause plenty of chaos behind the scenes.

In a blog post, former OpenAI engineer Calvin French-Owen, who helped build the company’s new coding agent Codex, said there “wasn’t any personal drama in my decision to leave” — but he did recap his experience at the AI company in a way that painted a picture of corporate bedlam as the company grows at extraordinary speed while keeping up with a steady stream of product releases.

That shouldn’t exactly come as much of a surprise, given the enormous stakes. The ongoing AI race shows no signs of letting up, with companies continuing to pour billions of dollars into expanding infrastructure and poaching top talent from competitors.

OpenAI’s trajectory, from a relatively quiet nonprofit to the frontrunner in one of the fastest-growing tech industries in just a few years, has caused plenty of disorder.

French-Owen recalled how “nearly everyone in leadership is doing a drastically different job than they were ~2-3 years ago” due to the explosive growth in headcount at the firm.

“Of course, everything breaks when you scale that quickly: how to communicate as a company, the reporting structures, how to ship product, how to manage and organize people, the hiring processes, etc,” French-Owen wrote.

While workers were encouraged to simply “just do things,” it was hard to keep up as “OpenAI changes direction on a dime,” reminiscent of Meta CEO Mark Zuckerberg’s infamous “move-fast-and-break-things ethos.”

“The company makes decisions quickly, and when deciding to pursue a direction, goes all in,” French-Owen wrote.

But setting a clearly-focused direction, particularly when it comes to code, has remained a pain point.

“Rather than having some central architecture or planning committee, decisions are typically made by whichever team plans to do the work,” French-Owen recalled. “The result is that there’s a strong bias for action, and often a number of duplicate parts of the codebase. I must’ve seen half a dozen libraries for things like queue management or agent loops.”

Interestingly, Elon Musk’s social media platform X-formerly-Twitter played a considerable role.

“The company pays a lot of attention to Twitter,” French-Owen wrote. “If you tweet something related to OpenAI that goes viral, chances are good someone will read about it and consider it.”

In short, the engineer paints a fascinating picture of what the ChatGPT maker looks like behind the scenes. With this much money on the line, the company has turned into a “very secretive’ and “more serious place” to work than before — don’t forget its recent turn into heavy security, including fingerprint scanners and more — since the “stakes feel really high,” per the former engineer.

But despite its enormous size, the multibillion-dollar company “still has that launching spirit,” French-Owen argued. As for what that means for everybody whose life and career is affected by its tech? They’ll have to watch from the outside.

More on OpenAI: Zuckerberg Is So Desperate for AI Power That He’s Building Temporary Data Centers in Tents