NTT DATA Group Codex Implementation and Results

NTT DATA Group Codex Implementation and Results

NTT DATA Group reduces incident analysis time from three days to 30 minutes

NTT DATA Group has successfully integrated OpenAI's Codex to automate complex technical tasks, most notably reducing a critical system incident analysis that previously required five experienced engineers and three days of work to just 30 minutes. This shift is part of a larger organizational transformation moving from a headcount-based revenue model to one focused on AI-driven value creation.

Foundation: Companywide ChatGPT Enterprise Adoption

NTT DATA Group established a global strategic partnership with OpenAI in May 2025 and deployed ChatGPT Enterprise across the organization. To ensure successful adoption, the company created an internal OpenAI Center of Excellence (CoE) responsible for license distribution, technical validation, usage monitoring, and the development of knowledge resources.

Internal survey data indicates high adoption rates and satisfaction:

  • Over 96% of respondents reported satisfaction with ChatGPT Enterprise.
  • Over 95% of respondents reported productivity gains.

This initial deployment built the necessary AI literacy and habits among employees, preparing them to transition from conversational AI to task-oriented automation with Codex.

Agentic Capabilities of Codex in Technical and Nontechnical Roles

Unlike standard coding assistants, Codex is utilized by NTT DATA Group as an agentic tool capable of independently investigating, executing, testing, and revising work based on instructions. This capability is being applied across two primary employee segments:

Technical Applications

Beyond the initial incident analysis success, the company has automated internal system operations using Playwright, reducing time spent on routine daily tasks. These automations are packaged as "Skills" to enable organization-wide adoption.

Nontechnical Applications

Under a "Client Zero" approach—where the company treats its own organization as the first customer—Codex is now used by nontechnical employees for tasks including:

  • Data Processing: Extracting transportation expenses from credit card statements into travel expense forms by understanding sheet structures and entry rules.
  • File Management: Organizing large volumes of files and analyzing data directly in Excel.
  • ** Lemmas and Reporting:** Creating analytical reports from raw data, reducing the need for specialized business intelligence tools and engineering support.
  • Process Automation: Scripting repetitive processes and summarizing documents.

Security and Governance Framework

To scale Codex to approximately 9,000 employees, the OpenAI CoE developed a security and governance framework. This framework provides practical guidance on:

  • Data usage permissions and system connectivity.
  • Network traffic management and sandbox mode application.
  • Appropriate levels of automation and the requirements for human review.

Organizational Lessons for AI Integration

NTT DATA Group identifies five key lessons for embedding agentic AI across an enterprise:

  1. Habit Formation: Integrate ChatGPT Enterprise into daily workflows to build AI collaboration habits.
  2. Network Effects: Broad deployment encourages peer learning and word-of-mouth adoption.
  3. Safety First: Establish a secure environment with robust security and privacy protections.
  4. Continuous Improvement: Use usage data, surveys, and employee interviews to refine adoption programs.
  5. Generalization: Use the CoE to identify high-impact use cases and convert them into reusable best practices with safeguards.

Summary of Tangible Results

  • Incident Analysis: Reduced from 3 days (5 engineers) to 30 minutes.
  • User Growth: Weekly active Codex users increased by 1.4x following the release of usage guides and hands-on training.
  • Operational Efficiency: Routine daily tasks reduced through Playwright-based automation packaged as reusable Skills.

Sources