Oracle, a cloud and database software company, is using ChatGPT Work for recruiting research and Codex for internal reporting and production engineering, reducing the time employees say they spend gathering information and completing routine tasks.
OpenAI detailed the deployments, reporting 130,000 active ChatGPT users and more than 95,000 active Codex users at Oracle.
Oracle’s talent acquisition team built a tool using ChatGPT Work that takes a job description and researches comparable roles, compensation benchmarks and talent pools across different locations.
Compiling this information previously took two to four days. Recruiters now spend about 15 to 20 minutes preparing for hiring-manager conversations using the tool, according to Jan Ackerman, Oracle’s senior vice president and global head of talent acquisition.
The tool has also made the recruiter intake process more consistent, Ackerman said. Previously, recruiters approached hiring research differently from one search to another. Hiring managers now receive the same type of market information regardless of which recruiter handles the request.
Business questions become database queries
Oracle Applications Lab, which helps run many of the company’s core business processes, has built a separate tool for internal reporting. The team built an ontology, a structured map of business objects, their relationships and rules.
Codex uses this information to turn questions written in everyday language into database queries, retrieve information from internal systems and generate reports or analyses.
“Business users now describe the outcome they want instead of hunting for a report,” said Richard Lam, group vice president of Oracle Applications Lab.
Lam described one employee who received an almost immediate answer to a question that previously took her a couple of hours to resolve. She then checked the AI-generated result against her usual manual process and found that the numbers matched exactly.
Oracle is also using Codex in production engineering. Site reliability engineers use it to gather information about incidents and retrieve the relevant response playbooks. According to Lam, what he described as a typical simple incident that previously took about an hour to resolve can now be handled in minutes, with engineers continuing to guide the response.
Employees remain responsible for the results
Despite the reported time savings, Oracle executives emphasized that the tools still require human oversight. Lam said employees remained responsible for system design, security and code maintainability. Barry Shilmover, vice president and technical adviser to Oracle’s CIO, described using Codex to turn ideas into prototypes rather than first putting them on paper.
In a separate technical post, Oracle said its enterprise ChatGPT connections preserve employees’ existing data permissions. It also described temporary privileged access and human approval for sensitive actions in its Codex and operations-agent rollout.