Toothpaste. Hand soap. Kitchen cleaners.
You likely have at least one Lion product in your bathroom or kitchen right now.
They are a classic, long-established manufacturer we rely on daily.
What would you think if I told you that this seemingly traditional company is currently one of the most advanced in Japan regarding the use of Generative AI?
Honestly, it’s surprising, isn't it?
Lion is a company with over 130 years of history. Normally, you'd think an old-school firm would be the furthest from AI. Yet, they are at the cutting edge.
This story was shared by Norihiko Nakabayashi (Executive Officer overseeing digital strategy) on the PIVOT program "&Questions," and it was incredibly fascinating.
The most impressive part? Within the company, the time spent searching through decades of research data was reduced to one-fifth.
This article breaks down that system into a format you can replicate in your own company.
The Problem: We Installed AI, but No One Uses It
Let me touch on a painful point first.
Has your company introduced ChatGPT or similar tools? Many companies find that only a few tech-savvy people use it. They did the training. They did the PoC (Proof of Concept). But it didn't lead to results and just faded away.
In the industry, this is sometimes called "PoC Death." It means dying at the PoC stage. It stops at "It seems convenient, but my work hasn't changed."
I think many companies feel the sting of that reality.
What makes Lion amazing is that they moved past this. They didn't just distribute AI; they created a system where frontline employees can "create AI themselves." This is why this article is relevant to everyone.
Searching Decades of Research Data Instantly: Search Time Reduced to 1/5th
First, let's talk about the strongest result.
Lion's researchers used to spend a lot of time asking, "Wait, where was that?" Research reports are text-heavy PDFs scattered across multiple folders. To find necessary information, they had to open folder after folder and summarize it themselves. Just imagining it is exhausting.
So, in 2024, they created a chat tool specialized for the R&D department. They put decades of research data—reports, product compositions, quality evaluations—into one place.
Now, when they ask the internal chat, it finds relevant documents and even summarizes multiple sources. "Searching" turned into "just asking." As a result, search speed and efficiency improved fivefold.
This might seem minor, but it's huge. R&D is the core of Lion. Speeding up research means speeding up the company's evolution. "AI for efficiency" is a tired phrase, but "researchers who spent all day searching folders now just ask a bot" is a concrete, powerful image.
The Foundation: Creating Lion AI Chat and Reaching 20,000 Uses per Week
This search tool didn't appear out of nowhere. There was a sequence. Before that, there was a very important, albeit unglamorous, foundation.
About three years ago, Lion rebuilt its entire company data infrastructure from scratch. Why? Nakabayashi-san said that without a radical overhaul, introducing Generative AI would start small and end small.
Many large Japanese firms have "siloed" data and systems that don't link in real-time. If the foundation is shaky, adding the latest AI will only yield superficial results. Investment in invisible areas pays off later.
Lion's digital strategy has three pillars: 1. Management infrastructure, 2. AI democratization, and 3. New business expansion. This article focuses on the second: democratization.
When ChatGPT was announced in November 2022, Lion moved fast. They internalized an internal chat tool called "Lion AI Chat" by Spring 2023. But tools alone aren't enough. Lion focused on adoption. They held hands-on sessions, gathered ideas through "Ideathons," and built communities on Slack and Teams. A tool is a "point," but a community is a "line" that leads to retention.
Lion AI Chat is smart; it switches between OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude depending on the task. They also don't store history, lowering the psychological hurdle for even executives to ask critical questions. Now, it's used 20,000 times a week and growing.
The Main Point: Democratizing the "Creator" Side
When the R&D tool was released, other departments immediately wanted in. Back-office wanted to search internal regulations; marketing wanted to search market share data. Everyone wanted AI to handle their tedious tasks.
But the expert AI team has limits. They can't build everything for everyone. Lion's solution? "Here are the tools; build it yourselves."
This is the "Democratization of the Creator Side." They chose a no-code tool called "Dify." It allows people without coding skills to build AI agents—small assistants that handle specific tasks. These are shared in an "Agent Hub," an internal marketplace where the best tools rise to the top. This creates massive motivation.
Preventing PoC Death: Training 100 People via the "Dojo"
How did they train these creators? They defined three types of talent: 1. Digital Entry (all staff), 2. IT Digital (experts), and 3. Digital Utilization (hybrid talent in sales/marketing/HR). They focused on the third group—people who know the business and can bridge the gap to technology.
They diagnosed 3,000 employees' skills using a radar chart to visualize gaps. Then they launched a "Dojo" (training camp). It wasn't just lectures; participants had to bring a real workplace problem and prove the ROI. If the business impact was low, the instructors were blunt. This rigor ensures that projects don't die after the training ends.
This year, they aimed to create 100 no-code AI creators. The 2-month training involved inventorying tasks, picking one suitable for AI, mapping the workflow, and building the app in Dify. Over 90% of participants continue to use Dify after the training because it actually made their jobs easier.
The Goal: Not Just Saving Time, but Transformation
Lion's goal isn't just reducing labor hours; it's "Smart Work"—transforming how they work to create higher value. As the workforce shrinks, they need everyone to be able to use and create AI to maintain productivity.
They are even building proprietary AI models. While general AI gives general answers, a model trained on Lion's data can suggest specific patented ingredients with evidence. They are also exploring digitizing "tacit knowledge" from veterans by analyzing videos of OJT sessions.
Conclusion: What is the Problem You Are Solving?
Nakabayashi-san emphasizes: "DX and digital are just tools." The most important thing is what you solve with them. Lion's "Vision 2030" aims for a "resilient profitability," and every AI initiative is connected to this goal.
Three takeaways for your company:
- Don't stop at "everyone using AI"; aim for "everyone creating AI."
- Training must be problem-driven with a focus on ROI to avoid "PoC Death."
- Invest in "hybrid talent" who understand both the frontline and the technology.
Lion is a toothpaste company, but they are using AI to change their culture and the company itself. Is your company just increasing users, or are you empowering creators?
