The key to results: "purpose-built generative AI." The key to adoption: "operational design."
"Honestly, I didn't know where to start." These are the words of a staff member at a third-sector company that adopted mitsumonoAI through the Fukui Tourism DX pilot. From September 2025 to January 2026, basicmath participated as a member of the Fukui Tourism DX Consortium, providing the generative AI suite "mitsumonoAI" as part of the Japan Tourism Agency's FY2025 "Tourism DX Regional Revitalization Model Pilot Project."
Why Fukui's Tourism Sector Needed Generative AI
The Fukui Tourism DX Consortium's initiatives have been selected for Japan Tourism Agency programs four consecutive years. At the November 2024 G7 Tourism Ministers' Meeting, Fukui's Tourism DX was featured as a leading case study in the outcome document "AI and Tourism."
Fukui Prefecture had accumulated valuable data — accommodation booking trends, tourist survey results — in its proprietary DMP platform "FTAS." However, a lack of expertise, dedicated staff shortages, and a disconnect between analysis and practical operations meant most field operators were unable to fully leverage this data.
The Three Walls Operators Faced
Wall 1: Time and Resources
Image creation requiring specialist skills, time-consuming review responses, complex manual summarization. A railway operator needed a full week for thumbnail creation; accommodation operators spent 10–15 minutes per review response.
Wall 2: The 0-to-1 Idea Gap
New menu development, event planning, PR campaigns — first steps stalled as ideas dried up and perspectives narrowed.
Starting ideas from scratch took time, and our perspectives tended to be one-sided.
Wall 3: The Objectivity Gap
Uncertainty about whether one's thinking is sound, a desire for objective feedback, no discussion partner — isolation in decision-making.
When I share my opinion, sometimes it gets dismissed as 'just a personal impression' and the discussion stalls.
mitsumonoAI's Design Philosophy
The top priority in developing mitsumonoAI was ensuring that "even AI beginners, without knowing any technical jargon, can use it immediately." Instead of requiring prompt engineering skills, we built purpose-driven agents and workflows where users simply fill in guided input fields to generate high-quality output.
Powered by multiple cutting-edge models — OpenAI, Claude, Gemini, Grok, and Perplexity — mitsumonoAI offers 35+ specialized AI agents and workflows across four categories: content creation, ideation, analysis and research, and documentation.
Tourism DX Advisor: "Sensei AI Satake Masanori"
A standout feature was "Sensei AI Satake Masanori," built on the expertise of the Fukui Tourism Federation's Regional Tourism Manager. Expert consultation was embedded as a product feature, enabling anytime access to strategic guidance without scheduling constraints.
The design focused on reproducing not just knowledge, but the personality, values, and communication style of the expert — delivering empathetic, context-aware responses rather than generic advice.
Early Adoption Challenges — Making AI "Personal"
The goal was to make AI a daily workflow "partner." But immediately after launch, operators hit three barriers:
1. The Connection Gap — How does AI relate to my actual work?
There are so many features that I struggled to decide which AI function to use for my specific task.
2. The Dialogue Gap — How do I communicate my intent to AI?
When I asked the AI to make corrections, it changed parts I didn't intend. I wanted to understand why.
3. The Application Gap — How do I turn AI output into action?
The agent function generates results, but I'm unclear on how to use those results to move my work forward.
Operational Design for Field Adoption
"Providing the system alone doesn't drive adoption." While registered users far exceeded the 100-user target, sustained usage was limited. We combined tiered learning, individual coaching, and ongoing support beyond basic tutorials.
Initiative 1: Tiered Seminar Program
We organized beginner, intermediate, and advanced seminars — 41 sessions in total — matching participants' skill levels to eliminate "I don't understand, so I won't use it."
Initiative 2: Individualized Coaching
Fully customized one-on-one sessions for 13 companies, tailored by industry: review response coaching for hotels, image generation + SNS workflows for railways, SWOT analysis for municipal staff.
The individual sessions provided practical, directly applicable information that was very valuable.
Initiative 3: Phone-Based Accompaniment
Regular calls focused not just on tool usage, but on uncovering latent challenges and co-creating AI-powered solutions.
Initiative 4: Live Chat Support
In-product chat support for immediate assistance, keeping operators from getting stuck.
Initiative 5: Ongoing Communication
Regular newsletters, blog posts with practical use cases, and shared seminar materials to build a self-sustaining learning environment.
Initiative 6: Continuous Improvement from User Feedback
UI/UX simplification, new agents, and feature refinements based on field input. Operators saw their suggestions reflected in the product, reinforcing continued engagement.
Results — Dramatic Efficiency Gains by the Numbers
Time Reduction
Thumbnail creation: 1 week
↓30 minutes
A railway operator with zero design skills could instantly generate high-quality images via mitsumonoAI. Time reduction: over 99%.
It feels like roughly 100,000 yen worth of value.
Time Reduction
Review responses: 10–15 min each
↓Cut by 50–67%
AI-generated draft responses in seconds, requiring only light editing. The psychological burden of responding to harsh reviews was also significantly reduced.
Time Reduction
PR scenario creation: half a day
↓1 hour
Freed approximately 10 hours per month — equivalent to ~20,000 yen in cost savings.
Time Reduction
Legal research: over 1 full day
↓~1 hour
From "I don't even know who to ask" to organized key points in about an hour — also reducing back-and-forth with legal counsel.
Beyond Numbers — A Mindset Shift
The most striking outcome wasn't time saved — it was that the way people approached high-value work fundamentally changed.
My perspective broadened. I started noticing that the root cause of issues was often somewhere I hadn't considered.
Objective feedback from AI gave me confidence in my own opinions.
AI reduces dependence on individual skills, preventing knowledge silos and raising the overall organizational capability.
AI has become a discussion partner — a 'good partner' in the truest sense.
Mindset Shift
From "individual experimentation"
to "organizational strategy"
Generative AI and "Omotenashi"
We believe "AI doesn't replace hospitality — it creates the time for people to focus on genuine hospitality."
By automating review responses, document preparation, and data analysis, generative AI frees operators to reinvest that time in face-to-face guest interaction and creating more compelling tourism experiences — work that only humans can do. This is the most practical and effective approach to structural challenges like labor shortages and aging populations.
The Fukui Model: 3 Steps to Successful AI Adoption
Step 1: Clarify the Purpose
Set specific, measurable goals led by management. Vague expectations lead to confusion and unused tools.
Step 2: Secure Trusted Guidance and Learning Opportunities
Tiered seminars for foundational knowledge, individualized coaching for specific challenges. An always-available support partner lowers psychological barriers.
Step 3: Scale Small Wins Across the Organization
Start with a single task, build confidence through quick successes, then share those wins organization-wide. Transform AI from an individual skill into organizational culture.
Field Voices Build the Future
The candid feedback, concrete improvement suggestions, and courage to embrace new technology from every participating operator — these are the driving force behind mitsumonoAI's evolution.
Key Takeaway
Generative AI is not just
for advanced enterprises.
With the right design and operations,
it takes root in any industry.
mitsumonoAI aims to be a partner in building AI adoption that truly takes root in the field.
What to delegate to AI. Where humans create value.
The more precisely you draw that line, the higher the quality of service.
That's the reality Fukui's pilot revealed.