NTT DATA case: IT incident analysis cut from hours to 30 minutes with AI

NTT DATA, one of the world's largest IT providers, shared the result of bringing AI into support operations: incident analysis time dropped from several hours to roughly 30 minutes. The model pulls data from logs and monitoring systems, reconstructs the timeline, spots anomalies and proposes a root-cause hypothesis — the engineer verifies the conclusion and makes the call.

The division of roles is the point: AI does the bulky, boring part — sifting thousands of log lines and matching timestamps — while judgment stays with the human. That split is what yields multiple-fold speedups without quality loss; attempts at "fully automatic" analysis, by contrast, often end in elegant but wrong conclusions.

The pattern travels far beyond IT: every post-mortem in a company has the same anatomy. A delivery slipped, a client left, cash flow broke — in each case you gather facts from correspondence and systems, build a timeline and find the point where things went wrong. If those reviews eat days of meetings at your company, they're a ready candidate for the same "AI gathers, human decides" pairing.