OpenAI’s engineers have cut incident investigation from hours to minutes using Codex Ultrafast, a high-speed mode in the company’s coding agent that has become the default tool for incident response inside OpenAI. The company shared the internal example as part of a series on how it uses its own technology.
OpenAI Codex Ultrafast (Credits: OpenAI)
When something breaks, engineers often do not have time to wait minutes for an answer while the issue compounds. With Ultrafast, they can investigate multiple potential causes at once and get answers without splitting their attention — a change that has reshaped how the company’s incident response channels actually work.
How OpenAI Uses Codex Ultrafast for Incident Response
Aaron Friel, a Member of the Technical Staff at OpenAI, is usually the one running point on an incident. He spins up a Slack channel for incident response and then points Codex at the problem, asking it to monitor, investigate, and answer questions as they come in.
“In any incident, there are usually many hops along the way of what could have gone wrong, and finding the right needle in the right haystack is really the chief job of the people on that channel,” Friel says. “Using Ultrafast, I’m able to essentially search all of those haystacks at the same time, and respond to incidents much faster.”
For security investigations, the time savings are concrete. AJ Trbojevic, a Datacenter Security Architect, says investigations used to take an hour or two to pull structured and unstructured data together, normalize it, and provide context for a decision. “Now that happens in 10 to 15 minutes,” he says. “And at times it’s near real time.”
Ultrafast Changes How Engineers Work Beyond Incidents
Adam Perry, who works on the core agent team building the code that lets Codex take actions and run commands, says Ultrafast changed his habits during security incidents. Before, he would ask Codex to look into something and then investigate it himself at the same time, because waiting was not fast enough. Now, he says, Codex is fast enough that he can just stay focused instead of splitting his attention.
His colleague Sayan Sisodiya uses Codex to dig through Slack for references to changes that might be suspect, or people talking about similar symptoms elsewhere — not just logs. And Jeff Niu, another Member of the Technical Staff, says most of his work now comes back in under 30 seconds, fast enough that the conversation happens in real time instead of breaking his train of thought.
Ultrafast is also changing bigger engineering work. Niu says he is now more willing to tackle large refactors, because reworking existing code takes far less time. “Technical debt is much easier to pay down now than it ever was before,” he says. “Code is cheap, and now it’s very quick to produce.”
The example is a window into how OpenAI itself builds with its own tools — and a sign of where AI-assisted engineering is headed: not just faster answers, but a workflow where the engineer never has to leave the problem to wait for the machine.
Also Read: Cloudflare Links Vulnerability Discovery to Defense With OpenAI’s Daybreak and Codex | OpenAI Models Are Coming to Google Cloud’s Model Garden
Source: OpenAI — read the full story there.
