Advancing organizational transformation for business innovation
OpenAI
DNP rolled out ChatGPT Enterprise across ten departments and patent research got 95% faster in three months. That's not a pilot program anymore, that's a company rewiring how it works.
Based on reporting by OpenAI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Dai Nippon Printing, the Japanese conglomerate better known as DNP, just published one of the more concrete enterprise AI case studies to come out of OpenAI's stable of customer stories. Instead of a single team quietly experimenting with a chatbot, DNP pushed ChatGPT Enterprise into ten core departments at once and tracked what happened over three months.
The numbers are specific enough to matter. Patent research, historically a slog of manual prior-art searches and legal cross-referencing, sped up by 95%. Overall processing volume across the deployed teams jumped tenfold. DNP says 87% of certain workflows are now automated, and 70% of knowledge produced in one part of the business is getting reused elsewhere instead of being reinvented from scratch.
That last figure is the one worth sitting with. Most large companies do not have a technology problem when it comes to institutional knowledge, they have a retrieval problem. Reports, research, and patent filings pile up in departmental silos, and nobody outside that team ever finds them again. If DNP genuinely lifted knowledge reuse to 70%, that suggests ChatGPT Enterprise is functioning less like a writing assistant and more like a company-wide search layer that finally makes past work legible to people who did not produce it.
DNP is a sprawling operation, with businesses spanning packaging, electronics materials, and printing technology, so a ten-department rollout is not a token gesture. It signals a company willing to standardize a single AI tool across genuinely different workflows rather than let each division bolt on its own point solution. Whether the 10x processing gain holds up as a durable productivity number or reflects a one-time catch-up effect from automating backlog is the part outsiders cannot verify from a vendor case study alone.
Still, this reads as one of the clearer examples yet of an old-line manufacturing and printing company treating generative AI as core infrastructure rather than a side experiment, and that shift in ambition is the real story here.
My take — AI-written commentary, not fact-checked reporting
Vendor-published case studies always deserve a raised eyebrow, since OpenAI picked this example precisely because the numbers flatter it. But even discounting for cherry-picking, a 95% cut in patent research time is the kind of claim that should make every legal and R&D department paying for enterprise AI ask why their own numbers look nothing like it. My bet is the gap is less about the model and more about how seriously a company bothers to restructure workflows around it.
Read more about this at: OpenAI