Dai Nippon Printing (DNP), one of the world’s largest printing companies founded in 1876, turned artificial intelligence into an everyday tool for its operation. In 2025 they deployed ChatGPT Enterprise across ten key areas and the results are clear: higher productivity, fewer manual tasks and, above all, preservation of institutional knowledge.
Sounds big — but what does that mean for you day to day? Think of it like turning tangled paper manuals and silos of expertise into something anyone in the company can query, like asking a colleague for a quick tip.
What DNP did and why
The adoption wasn’t improvised. In April 2023 they decided to integrate AI across the organization; in May they set up a secure environment for enterprise use; and in February 2025 they launched ChatGPT Enterprise in ten priority departments.
They set concrete goals: every employee had to use the tool at least 100 times per week and achieve more than 50% automation in time reductions. Why so ambitious? Because they aimed for measurable, scalable results — not isolated experiments.
Measurable results
The numbers are striking and easy to understand:
- 90% of use cases showed measurable results.
- 100% weekly active use in the adopting areas.
- 87% automation rate in time reduction.
- 70% reuse of knowledge through
custom GPTs. - 10x increase in the volume of processes handled.
Those percentages aren’t promises; they’re operational impacts that changed how people work every day.
Use cases that matter
Patents and intellectual property strategy
The innovation team automated patent searches, summaries and classification, cutting research time by 95% and increasing coverage by 10x. That helped reduce rejections and reversals in filings by spotting key differences against competitors.
They drove adoption by making usage visible, sharing learnings and iterating, which created scalable impact.
Research and development
In R&D, complex tasks that used to take months or years were implemented in days. Concrete wins:
- Structuring patent information and equipment principles in 3 days, instead of several months.
- Employees without Python experience generated and ran code with help from
ChatGPT Enterprise. - Development work that would have taken over a year was completed in a few days.
That freed up time for human expertise to combine with automation and for ideas to emerge faster.
IT governance and security
Audit and review processes sped up noticeably:
- External audit comparisons from 30 to 5 minutes.
- Selection of cryptographic suites from 3 to 1 hour.
- Initial review of ~100 CIS noncompliance items in 10 minutes instead of two workdays.
The model helps gather relevant data and produce clear outputs, though final verification remains human.
Knowledge preservation, the core challenge
DNP’s experience shows the biggest risk is losing know‑how when experts leave or information lives on paper. Their response was to turn those records into structured, accessible data via custom GPTs.
- Data architecture definition time reduced by 90%.
- They doubled the number of technical papers they could review.
Their goal is to transform generational knowledge into digital work, mitigating labor shortages and ensuring continuity of expertise.
What you can learn and apply
Sound like a scalable plan? Some practical lessons:
- Make usage visible: clear metrics accelerate adoption.
- Empower non‑IT people: the tool should lower the technical barrier.
- Document so the AI can learn: structuring data is an investment in longevity.
- Create reusable templates:
custom GPTslet you share best practices. - Keep human control: AI helps decide, but final verification is people’s job.
Final reflection
DNP didn’t just chase efficiency; they sought to turn tacit knowledge into lasting competitive advantage. By structuring manuals, logs and experience into data the AI can consume, they prepare the company for a smaller workforce and an operation where innovation rests on repeatable processes.
Can you imagine what your team could do if part of the know‑how were available to everyone in the company with a single question?