Hanayama Co., Ltd. conducted a six-month AI-generated training program called "Corporate Reskilling" for employees from various departments including accounting/finance, sales, planning/marketing, research/development, and administration. Based on the responses of the 10 participants who completed the survey, the amount of time saved by AI was...From approximately 1,176 hours to approximately 1,560 hours per year.This resulted in an increase of approximately 331 TP3T. The number of tasks using AI has also expanded from 62 to 81 for the 10 people involved.

At the beginning of the training, there was a clear difference between members who used AI daily and those who hadn't yet touched it. From there, everyone progressed to the stage of using AI in their own work. Some members even started creating their own business applications. We will introduce how this puzzle and game manufacturer, founded in 1933, expanded its use in just six months.

Company Introduction

Hanayama Co., Ltd., founded in 1933, is a specialist manufacturer of puzzles and games headquartered in Chiyoda-ku, Tokyo. They have been manufacturing and selling toys such as family games, 3D puzzles, and bingo since the early Showa era, supplying products not only to domestic markets but also to overseas markets. They operate a logistics center and customer service center in Ichikawa City, Chiba Prefecture, and conduct business in collaboration with group companies that handle logistics, OEM, overseas manufacturing and sales, and packaging.

The company's flagship product is the disassembly puzzle."Huzzle"The only rules are "Take it apart. Put it back together." This series, featuring designs by some of the world's leading puzzle designers brought to life with the company's precision casting technology, boasts over 50 designs that are popular worldwide as puzzles for adults. Other series include "Katsunou" and "Logical Newton." In 2026, two of their products received Excellence Awards at the "Japan Toy Awards 2026," and they were also selected as one of the "20 Toy Industry Companies Opening Up a New World" by the Ministry of Economy, Trade and Industry.

The message we are promoting is"Making the unknown interesting."The company's characteristic attitude—that the unknown is what makes it interesting—was also evident in their approach to generative AI, a new tool.

Issues before introducing generative AI

The training included employees from various job roles. Their daily tasks differed greatly, encompassing accounting/finance, sales, planning/marketing, research/development, and administration. Therefore, we began by having each participant take stock of their own tasks and identify where their time was being spent.

The inventory process identified 105 tasks, totaling approximately 192 hours per week. The largest portion of these tasks involved writing documents and materials, such as monthly reports and proposals for retailers and wholesalers. These tasks alone accounted for over 30 hours per week. Checking purchase orders took about 4 hours per week, and data entry and formatting took about 5 hours per week; AI was hardly being used for these verification and formatting tasks.

There were also differences in how AI was used. At the beginning of the training (the first workshop), 7 out of 10 participants were already using AI 4 days or more per week. On the other hand, some participants were too busy with their regular work to use AI and reported using it 0 days a week and for 0 tasks. There was also a big difference in the number of tasks that people were using AI for. In other words, it was a polarization: some people could use it, and others could not.

Implementation details - Hanayama's unique initiatives

The program, which lasted approximately six months, was designed with the goal of enabling everyone to use AI in their own work, and the content was adjusted based on feedback from the participants.

• Building a foundation through video lessons and four joint workshops.
  After explaining how to use the tools and learning system at the startup meeting, we began over 10 hours of video learning. The joint workshop consisted of four sessions, covering the basics of prompts and file utilization, creating Gems (chatbots), and two sessions on Google Workspace x Gemini. The goal is to enable the use of AI within the Google tools that you use every day.
• Based on feedback from participants, we have revised the content of the support meetings for practical application.
  During the interim report meeting, it was noted that beginners were struggling to keep up with the amount of information and the pace of the session. On the other hand, it was also revealed that many tasks involve the use of spreadsheets, highlighting a high need for automation. Therefore, subsequent support meetings were switched to a more focused, lecture-style format. In the group work rooms, the organizers now resend prompts as needed.
• Three support meetings using our company's operations as a case study.
  In the first session, we used Google Workspace Studio to work on automatically sorting inquiry emails and designing automated workflows for our own tasks. In the second session, we used Gemini to work on three tasks: creating new product proposal materials for retailers and wholesalers, checking 16 purchase orders, and formatting and summarizing order data. In the third session, we combined NotebookLM and Gemini to explore efficiency ideas for each department based on our company's documents. In every session, we shared the understanding that "AI is for preparation, but facts, numbers, and final decisions are made by humans."
• Tracking changes through business process inventory and surveys
  Before and after each training session, we collected information through questionnaires about the number of tasks participants were using AI for, how often they were using it, and the time they were able to save. We shared the changes observed after each session in interim and final reports.

Benefits of using generative AI

The biggest change was the total amount of time saved by AI. The combined total of the 10 participants who responded to the survey increased from 24.5 hours per week at the beginning of the training to 32.5 hours per week by the final response. When converted to an annual amount, this represents an increase from approximately 1,176 hours to approximately 1,560 hours.

Time saved by AI (total for 10 participants, calculated annually)(Based on participant survey responses)

BEFORE	Approximately 1,176 hours
AFTER	Approximately 1,560 hours
Rate of increase in time reduction	Approximately 33%
Tasks involving the use of AI (total for 10 people)	81 items
Participants who use AI in their work (out of 10 people)	10 people
Time saved per year	Approximately 1,560 hours(Total for 10 participants: 32.5 hours per week x 4 weeks x 12 months)

Across all participants, this amounts to an additional 384 hours per year. Not only has the amount of time saved increased, but the number of tasks using AI has also increased from 62 to 81 for the 10 participants combined, an increase of approximately 311 TP3T.

Changes in key KPIs (overall for 10 participants)

index	change	stretch
Tasks involving the use of AI (total for 10 people)	62 items → 81 items	Approximately 31% increase
Annual time savings (total for 10 people)	1,176 hours → 1,560 hours	Approximately 33% increase
Workshop satisfaction (average)	3.70 → 4.14	Approximately 12% increase

Workshop satisfaction initially dropped to 3.10 in the second session. After reviewing the operation, it rose to 4.14 in the fourth session, with 86% respondents giving it a rating of 4 or higher on a 5-point scale.

The extent to which AI is being used in business operations (number of participants out of 10)

Business area	Early stages of training	final answer
Data analysis and spreadsheets	4 people	7 people
Compiling and replying to emails	5 people	5 people
Creating meeting minutes and business negotiation records	5 people	5 people
Preparation of documents, proposals, and project plans.	4 people	5 people
Image generation and editing	3 people	4 people
translation	2 people	3 people

The most widespread use was in data analysis and spreadsheets, increasing from 4 to 7 users. Applications such as having AI create spreadsheet functions, organizing inventory data lists, and cross-referencing and analyzing multiple documents are spreading to sales, administration, and research and development. Half of the users were already using AI for email creation and meeting minutes from the beginning of training. Now, with the addition of document creation, image generation, and video creation, a model is emerging where AI is handling a large portion of "writing" tasks.

* The figures are based on responses from participants in the post-workshop questionnaires. The initial data is from the first workshop (one participant who did not respond uses their response from the second workshop), and the final data uses each participant's last response. The annual calculation is calculated as weekly time saved × 4 weeks × 12 months, and the results will change if the assumptions change.

These changes are also reflected in the free-response answers in the surveys. One participant who used AI only 0 days a week at the beginning of the training wrote the following in the session on using spreadsheets:

"By entering text, it creates a function and automatically inputs the data."

From participant surveys

In their final response, this participant began using AI to verify functions when organizing inventory data lists. Other participants have also shared how they are applying what they learned to their own work.

"Being able to use spreadsheets and Gemini simultaneously is helpful for my work."

From participant surveys

"I felt that this could be useful when creating project proposals and compiling documents within development work."

From participant surveys

There have also been further advancements. Some participants, inspired by their experience using Gemini during the training, have begun experimenting with other tools such as Claude Code, creating their own internal business tools, and automating repetitive tasks by registering them as "skills." Building upon the foundation of Gemini acquired during the training, they are now entering a stage where they choose their own tools and transform entire business processes.

Future outlook

Now that AI-savvy individuals are more widespread within the company, Hanayama's next challenge is to integrate individual ingenuity into organizational systems.

• Spread the skills of the project members throughout the company.
  We will make the business automation skills developed by the project members available to other members. We are considering ways to provide these skills in an easy-to-use format, such as via a desktop application, so that even members unfamiliar with terminals can use them.
• Expanding the automation of spreadsheet tasks
  We will gradually incorporate spreadsheet integration and automation using Workspace Studio, which were highly requested during training, into our daily report creation and verification tasks.
• Share use cases and encourage everyone to use them.
  We will promote the sharing of use cases and aim for everyone to be able to use AI in their own work.
• We are also considering using other AI tools.
  Based on the foundation I've built with Gemini, I'm also considering using other AI tools such as Claude.

In about six months, the generation AI went from being a tool used by only a few members to a tool used by all 10 members in their work. The reduction of approximately 1,560 hours per year is one of the results achieved along the way.

"Making the unknown fun." — Hanayama has been bringing a culture of enjoying the unknown through puzzles and games. Now, they are also embracing the new "unknown" that is generative AI and turning it into their own tool.