We are Shaping the Future! Showcasing Success Stories as We Innovate for a Sustainable Tomorrow
Recently, generative AI has become an essential tool for many companies, helping improve work efficiency, create content, and design products. Understanding its importance in driving digital transformation (DX) and business growth, OMRON started AIZAQ*1 in 2023. This companywide project encourages the voluntary use of generative AI, starting with its employees in Japan.
*1 "AIZAQ" stands for "AI with Zest, Accelerate and Quest," which is meant to evoke associations with Sir Isaac Newton and his wisdom and intellect as a scientist.
縲
While generative AI has been spreading fast among businesses overall, only a few leading companies are actively using it. According to the 2024 White Paper on Information and Communications in Japan published by the Ministry of Internal Affairs and Communications, Japan, 46.8% of Japanese companies use generative AI in business, which is much lower than in the U.S. (84.7%), China (84.4%), and Germany (72.7%).*2 The slow implementation in Japan is mainly due to challenges in knowledge sharing and the penetration of the technology within companies. Even when some employees proactively and effectively use generative AI, their successes rarely spread through the company because the details and benefits of these initiatives are not shared internally, often leading to temporary efforts.
OMRON was no exception to this. Despite recognizing the usefulness of generative AI, technical hurdles prevent its spread throughout the company, giving rise to gaps in how employees use generative AI.
*2 2024 White Paper on Information and Communications in Japan, Ministry of Internal Affairs and Communications, Japan
https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r06/html/nd151120.html (Ref. Nov. 7, 2024; in Japanese only)
縲
A member of the AIZAQ Project Management Office (PMO), Tatsuya Koga of the DX-Business Innovation Center, Innovation Exploring Initiative H.Q., recognized the need for better knowledge sharing. A key challenge was helping employees create relevant prompts, which may found difficult at first. Prompts are crucial for effectively using generative AI, as the quality of prompts directly affects the results users receive. By setting up a portal site called AIZAQ Canvas, Koga hopes to empower every employee to use generative AI effectively through knowledge sharing, ultimately driving company growth and improving competitiveness.
Left: Project Leader Koga (second from right) with his prompt library creation team / Right: Portal site AIZAQ Canvas
Various organizations, such as governments, enterprises, and relevant institutions publish their prompt libraries, but some prompts need adjustments to fit our specific company needs. Koga saw the need for an accessible prompt library that anyone can copy and paste. He initiated the creation of one with his team of about ten members, and releasing the first edition about three months after starting.
Before developing the prompt library, Koga set a clear goal: to create "prompts that make people want to use them." He believed that no matter how many prompts were created, they would be meaningless if employees didn't use them. With this in mind, he engaged his entire team in repeated discussions about what would motivate OMRON employees to actively use their prompts.
For prompts to be effective, they must meet user expectations. "Prompts fail if they consistently provide irrelevant responses," explains Koga. Users lose trust in prompts when they fail to meet expectations and are unlikely to use them again. For the same reason, employees may avoid accessing the prompt library altogether. Therefore, successful prompts must generate responses that closely align with users' expectations.
To create quality prompts, Koga's team conducted regular reviews and tested prompts in simulated work scenarios. This process continued until everyone on the team was satisfied with the outcomes.
Another aspect Koga valued was mastering the basic techniques of prompt creation. Since generative AI responds based on the prompts it receives, understanding and applying these techniques is essential.
The three basics of creating prompts
For example, clearly specifying roles, such as "#Role: {You are a customer support specialist}," increases the likelihood of receiving desired responses. Similarly, using positive instructions, such as "Make it easy for beginners to understand," is more effective than negative ones like "Do not use technical terms." Technical advisors who support the team with their expertise were invited to help the team learn these practical basics in creating such prompts.
縲
The responsibility of spreading new technologies across the company should not rest solely with experts; it should involve all employees. On AIZAQ Canvas, the prompt library has fostered a culture where employees share their experiences and achievements in using generative AI. It also serves as a guide for newcomers, answering questions like, "What is generative AI in the first place?", "What tasks is it suited for?", and "What should we be careful about when using it?" The future plan is to develop AIZAQ Canvas into a platform where employees unfamiliar with generative AI can gain insights into how to apply it in their work.
Going forward, as generative AI sees even broader applications, the knowledge accumulated through AIZAQ Canvas is expected to strengthen our competitive edge both internally and externally and inspire new innovations.
縲
It all started from Koga's burning desire to "make generative AI easily available to all employees, not just a few." His brainchild, the prompt library, gradually gained interest within the company, with initial traffic of 1600 quickly grew to 5600 within a month. Since the library not only provides prompts but also tips on their use, many employees reported successful experiences in achieving their desired outcomes. Hearing positive feedback and success stories from his peers, Koga realized that the prompt library, created in collaboration with his team, had significantly contributed to promoting the effective use of generative AI, the very goal of AIZAQ.
Knowledge sharing is one of the most effective ways to facilitate the greater use of generative AI. Introducing generative AI is just the beginning; sharing knowledge helps employees deepen their understanding and use it in ways that benefit the business. Starting in November 2024, some of the project themes on the effective use of generative AI will enter the phase of organizational implementation.
"With this prompt library, I aim to empower all OMRON employees to utilize generative AI. Through knowledge sharing, I hope to contribute to boosting DX within OMRON, enhancing customer value, and driving further innovation," says Koga. OMRON's efforts to promote the effective use of generative AI will continue.
