Oracle Deploys ChatGPT Work and Codex to Cut Multi‑Day Tasks to Minutes

TL;DR

Oracle has rolled out ChatGPT Work and Codex to more than 100,000 employees, turning tasks that previously required days of specialist effort into minute‑long operations across recruiting, business analytics, and site‑reliability engineering.

Recruiting – market intelligence in 15‑20 minutes

Outcome: Recruiters now generate comparable‑role research, compensation benchmarks, and talent‑pool assessments in under 20 minutes instead of 2–4 days.

  • The talent acquisition team built a talent market intelligence tool with ChatGPT Work.
  • The tool ingests a job description, automatically researches comparable roles, benchmarks pay, and evaluates talent pools across locations.
  • Jan Ackerman, SVP & Global Head of Talent Acquisition, reports that the process is now consistent and fast, enabling recruiters to prep for hiring‑manager conversations in 15‑20 minutes.

“We’ve built our talent market intelligence tool using ChatGPT Work, and it’s really revolutionized how we do the front end of our recruitment process.” – Jan Ackerman, Oracle

Business analytics – natural‑language queries become SQL instantly

Outcome: Business users describe desired outcomes in plain language and receive accurate SQL‑driven reports or applications within seconds.

  • Oracle Applications Lab created an ontology of corporate objects, relationships, and rules.
  • Codex translates a user’s natural‑language request into the appropriate internal system calls and generates the required SQL query.
  • A user reported receiving a correct answer “almost immediately,” matching the manual process down to the exact numbers.

Site‑Reliability Engineering – incident resolution in minutes

Outcome: Simple incidents that once took an hour to resolve are now handled in minutes through automated context gathering and playbook retrieval.

  • SREs invoke Codex to collect incident context and pull the correct run‑book.
  • Engineers spend more time decision‑making and less time information‑search.
  • Richard Lam, Group VP of Oracle Applications Lab, emphasizes that the shift is a “business transformation” rather than a fully autonomous system.

“Business users now describe the outcome they want instead of hunting for a report. That is a major shift. I would even call it a business transformation.” – Richard Lam

Leadership lessons for AI‑augmented development

Key take‑aways:

  1. Guardrails are essential. System design, architecture, and security must be defined before Codex generates code.
  2. Prototype‑first, spec‑later. Teams now start with a working prototype rather than a detailed specification.
  3. Own the generated code. Continuous human oversight is required to keep the codebase maintainable.

“If you don’t work alongside Codex, you’re going to get into a situation of having lots of code that is not going to be maintainable.” – Richard Lam

Future direction – AI as a universal work engine

Projection: Oracle is shifting the entire organization to outcome‑driven workflows where ChatGPT Work and Codex handle the implementation details.

  • Thousands of employees already rely on the two models for daily tasks.
  • Executives anticipate that future problems will be approached first by prototyping with Codex, then refining the solution.

“I don’t know what the answer is because I don’t know what my next problem to solve will be. I just know that one of the first things I’ll do is I’ll leverage Codex to do that.” – Barry Shilmover

Implications: Oracle’s internal deployment demonstrates that large‑scale enterprise adoption of LLM‑powered tooling can compress multi‑day workflows into minutes, improve consistency, and free specialist talent for higher‑value activities. The case study highlights the importance of strong governance, human‑in‑the‑loop oversight, and clear outcome‑driven prompts when scaling generative AI across an organization.

Sources