What changed
OpenAI’s October 8 customer story says Oracle has 130,000 active ChatGPT users and more than 95,000 active Codex users. Oracle reports that talent-acquisition research takes 98% less time with the workflow. It describes employees in recruiting, engineering and operations turning specialist work that could take days into repeatable tasks completed in minutes.
The examples include recruiters doing market research, business users describing an outcome instead of hunting for a report, and technical leads building internal tools with Codex. The practical shift is from asking for one answer to delegating a bounded workflow: gather context, create an artifact, inspect it and continue.
What the result actually shows
Oracle is the customer reporting these results and a commercial customer of OpenAI, so treat the figures as a company case study rather than an independent controlled benchmark. The article does not provide a measurement protocol for the 98% recruiting reduction.
For another organization, the useful test is specific: which handoffs disappeared, how much review remains, and can the time saved be compared with a measured baseline? That is a better indicator than counting prompts or generated lines of code.
The accompanying screenshot is from Oracle’s separate HCM AI-agent documentation; it illustrates an enterprise interface and is not the ChatGPT Work or Codex system in this case study.
Source published 2026-10-08. Coverage is based on the maker’s announcement and demonstration.