Skip to main content

NTT DATA's AI Agent Codex Cuts 3-Day Failure Analysis to 30 Minutes, 9,000 Employees Onboard

Imagine a team of five senior engineers spending three days troubleshooting a system failure. Now imagine that same task done in half an hour. That's the reality for NTT DATA Group, a Japanese IT services giant, after it deployed OpenAI's Codex agent.

OpenAI recently shared a case study highlighting how NTT DATA transformed its operations. Codex, an AI agent that can independently investigate, execute, test, and modify code, has become a game-changer. About 9,000 employees—from developers to finance staff—now use it daily.

From ChatGPT Enterprise to Codex: Building AI Habits First

NTT DATA's journey began in May 2025, when it formed a global strategic partnership with OpenAI. The company rolled out ChatGPT Enterprise across the organization and set up an internal OpenAI Excellence Center to drive AI adoption. The results were impressive: over 96% of employees reported satisfaction with ChatGPT Enterprise, and more than 95% felt their productivity had improved.

This phase wasn't just about tools—it was about culture. Employees got into the habit of using AI for research, writing, and content creation. That foundation made the introduction of Codex smoother. As Hiroaki Sato from NTT DATA's AI technology department put it, Codex changed the company's perception of AI. The idea that "AI can lead work completion" had an impact comparable to the arrival of ChatGPT itself.

Shoji Shono, head of the Global AI Office, emphasized that Codex isn't just helping developers write code faster. It's opening up a new way of working for every employee.

From Engineers to Non-Technical Staff: AI for Everyone

NTT DATA adopted a "Customer Zero" strategy—treating itself as the first customer for AI applications. Employees were encouraged to test Codex in their daily work. And they did.

Non-technical staff found creative uses: extracting transportation costs from credit card statements to fill out travel reimbursement forms, analyzing Excel data to generate reports, and batch organizing files. Tasks that once required engineer support can now be done independently.

This expansion didn't happen overnight. The Excellence Center developed detailed security guidelines, covering data use, system connection permissions, sandbox mode configuration, automation levels, and manual review requirements. After releasing a usage guide and conducting hands-on training, the weekly active Codex user base grew by 1.4 times.

Image

A Blueprint for Enterprise AI

NTT DATA's experience offers lessons for other organizations. First, build AI literacy before introducing advanced agents. Second, let employees experiment in safe environments. Third, invest in governance from the start.

The company's vision is to amplify the knowledge and professional skills of each employee through its "AI-Driven Enterprise" strategy. And it's not stopping internally—NTT DATA plans to share its successful internal experience with customers and society.

For now, the numbers speak for themselves: a 3-day task reduced to 30 minutes, 9,000 active users, and a workforce that's learning to work alongside AI agents. The future of enterprise AI might look a lot like NTT DATA's present.

Key Points

  • NTT DATA reduced a 3-day system failure analysis to 30 minutes using OpenAI's Codex agent.
  • About 9,000 employees now use Codex, including non-technical staff for tasks like expense reporting and data analysis.
  • The company's "Customer Zero" strategy and security-first approach enabled safe, widespread adoption.
  • Building AI habits with ChatGPT Enterprise first paved the way for more advanced agents.