Inside OpenAI’s race to catch up with Cloudcode

While some people describe her virtual persona as “dry bread,” many have come to appreciate her less fawning style, says Katie Shi, a researcher who works on Codex behavior at OpenAI. “A lot of engineering work is about being able to take critical feedback without interpreting it as mean,” Shi says.

Many major companies have signed on to use Codex as well. “The fact that ChatGPT is synonymous with AI gives us a huge advantage in the B2B market,” he says. Ransom SimoCEO of Applications at OpenAI. “Companies want to use technologies their workers already know.” OpenAI’s strategy for selling Codex is largely based on packaging it with ChatGPT and other OpenAI products, Simo said.

Jeetu Patel, Cisco’s president and chief product officer, says he told employees not to worry about the cost of using Codex, because they would need to feel comfortable using the tool. When employees ask if they’ll “lose their job because they’re using these tools,” Patel says, “What we should be saying to our employees is no, but I guarantee you’ll lose your job if you don’t use them, because you won’t be relevant. So you’ll be out of business.”

Today’s panic About AI coding agents has spread far beyond Silicon Valley. The Wall Street Journal credited Claude Code with causing the… $1 trillion in tech stock sell-offs Last month, as investors feared the software would become completely outdated. Weeks later, IBM stock had its worst day in 25 years after Anthropic announced that Claude Code could be used to update legacy systems running COBOL, common in IBM machines. OpenAI has worked tirelessly to make its AI coding agent part of the community conversation, spending millions of dollars on a Super Bowl commercial about Codex, rather than ChatGPT.

At Mission Bay Temple, no one needs to participate in the Codex. Many OpenAI engineers I spoke with said they rarely write code at all. They spend their days just talking to Codex. Sometimes they get together and do it in a group.

At headquarters, I participated in a Codex hackathon, where about 100 engineers crowded into a large room. Everyone had four hours to build the best demo using Codex. One of OpenAI’s senior leaders stood at the front of the room, looking away from the laptop in his hands and mouthing the team’s names into the microphone. Team representatives nervously walked to the podium and gave short speeches about their AI projects through trembling voices. Winners received Patagonia backpacks.

Many projects have been created using Codex and are designed to help engineers use Codex better. One group built a tool that summarizes Slack messages into weekly reports. Another group has built an AI-generated Wikipedia-style directory for OpenAI’s internal services. Many of these demonstrations would previously have taken days or weeks to begin, but now can be done in an afternoon.

On my way out the door, I bumped into Kevin Weil, the former Instagram executive who now heads OpenAI for Science, the company’s new unit for building AI products for researchers. He told me that Codex had been working for him on some project overnight, and that he would check on it in the morning. This has become standard practice for Weil and hundreds of other employees. One of OpenAI’s goals for 2026 is to develop an automated intern that will research (what else?) artificial intelligence.