Calvin French-Owen thought he had a sense of what he was getting himself into when he began at OpenAI in May of 2024. The MIT graduate and successful entrepreneur had co-founded Segment, a customer data company that Twilio acquired for $3.2 billion in 2020. But despite all of his Silicon Valley bona fides, there was one thing that didn’t quite ready him for the whirlwind of working at one of the world’s most advanced AI research facilities.
After a little more than a year at the ChatGPT-maker, French-Owen recently departed the company and wrote a lengthy blog post about the experience. His leaving wasn’t because of any office politics, he claimed, but because he felt compelled to go back to his startup roots.
“I wanted to give my take because there is so much hype and bluster about what OpenAI is doing, and very little ground-level information about what the culture of being there actually is like,” French-Owen posted on his blog.
Challenges, Culture, and the Meta Talent Drain
What French-Owen found at OpenAI was a company in its record-breaking expansion. The company expanded from 1,000 to 3,000 employees on a yearly basis, and there is a reason for that rate: ChatGPT was the fastest-growing consumer product in history with 500 million monthly active users, and it is continuing to grow.
But meteoric growth is accompanied by monumental challenges. French-Owen explained how fast scaling shatters everything from communication systems to product development processes. “Everything breaks when you scale that fast: how to communicate as a company, the reporting structures, how to ship product, how to manage and organize people, the hiring processes,” he described.
The disorganization of growth is also apparent in the everyday way the firm is run. French-Owen described one unorthodox element of OpenAI culture: total lack of email. “Everything, and I mean everything, is on Slack,” he said. “No email. I maybe got ~10 emails in my entire time there.”
French-Owen also mentioned what he referred to as a “very significant Meta → OpenAI pipeline” in engineering recruitment. That remark gains further light in the wake of recent news reports of Meta aggressively poaching talent from OpenAI with reportedly massive compensation packages, including a $100 million signing bonus that even OpenAI CEO Sam Altman called “crazy.”
The Intense Culture of OpenAI, High Pressure, Fast Pace, and Startup Life
OpenAI’s engineering culture is similar to early Meta in so many ways, French-Owen explained. There is a hit consumer application, infrastructure development, and extreme pressure to get things out the door.
Like most startups, individuals are encouraged to go after their ideas, although this can result in duplicated effort. French-Owen explained that he saw “half a dozen libraries for things like queue management or agent loops.”
The coding ability ranges from Google veterans to recently graduated PhDs with minimal real-world experience. Given that the company relies so heavily on Python, their main codebase can be “more of a dumping ground,” French-Owen added.
Arguably, the most telling aspect of French-Owen’s story was his account of working on releasing Codex, OpenAI’s programming tool based on AI. He explained it was one of the most difficult periods of work he’d ever done, working until 11 PM or midnight each night, waking up at 5:30 AM to catch a glimpse of his newborn, and getting to the office at 7 AM and working most weekends.
His team of around 17 members – eight engineers, four researchers, two designers, two go-to-market staff, and a product manager – built and released Codex in seven weeks. The effort paid off big time. “I’ve never seen a product get so much instant adoption just from appearing in a left-hand sidebar, but that’s the power of ChatGPT,” French-Owen said.
OpenAI Prioritizing Real-World Threats Amidst AGI Development
Pushing back on recent criticism of OpenAI’s strategy regarding AI safety, French-Owen came to the defense of the company against accusations that it is not taking safety seriously. The emphasis, he thinks, is more on real-world threats such as hate speech, abuse, and prompt injection than on speculative dangers such as intelligence explosion or power-seeking.
“That’s not to say that nobody is doing the latter; there are definitely people who are doing theoretical risks. But in my view, it’s not the priority,” he said, while conceding that OpenAI ought to release more of their safety research. While the work is demanding and the culture is intense, French-Owen painted OpenAI as a serious, mission-oriented company in which the stakes appear to be phenomenally high.
Meanwhile, the firm is attempting to create artificial general intelligence while also having a product that hundreds of millions of people depend on for everything from medical advice to therapy. This behind-the-scenes tour gives us a glimpse into one of the world’s most influential companies that will shape the direction of our technology future, and it reveals the excitement and exhaustion of developing tomorrow’s AI technologies.
