NTT DATA slashes 3-day incident analysis to 30 minutes using OpenAI's Codex

A tedious, multi-day forensic slog that tied up five senior engineers for three days now takes half an hour. That’s the headline number from NTT DATA Group after deploying OpenAI’s Codex to roughly 9,000 employees. The firm, a global IT services giant, had the engineers replicate a complex incident analysis they’d previously done manually. Codex completed it in 30 minutes—a 99.3% reduction in time. This wasn’t a minor productivity bump. It was a flashing neon sign, and senior leadership noticed immediately.

The speed is impressive, but the real story is who’s using the tool. Building on a broad rollout of ChatGPT Enterprise that left 95% of staff reporting productivity gains, NTT DATA didn’t restrict Codex to developers. Non-technical employees are now using its agentic capabilities to build their own lightweight tools, wrangle messy Excel data, and automate bureaucratic nightmares like extracting travel expenses from credit card statements. Work that previously required a favor from the engineering department is becoming self-service.

Hiroaki Sato from the company’s AI Technology Department captured the cultural shift, saying, “The idea that AI can take the lead in carrying out work has had an impact similar to the arrival of ChatGPT.” It’s a telling comparison. The first wave was about an AI copilot for brainstorming; this second wave is about delegating execution. Under a “Client Zero” strategy, NTT DATA is treating its own workforce as the first test case, and the results are creating momentum that’s pulling in non-engineers across the company who are finding their own messy, repetitive tasks to automate.

It’s a data point that cuts through a lot of AI hype. We’re not talking about a theoretical future of work. This is a concrete example of agentic AI moving from generating text to executing multi-step, analytical grunt work with minimal human hand-holding. The open question, which this case study doesn’t answer, is how the role of those five senior engineers evolves. Do they spend their newfound 35-hour windfall on higher-level architecture, or does the tool reshape the team structure entirely?