Key Takeaways (3-Point Summary)
・Business Overview: ICOMA Inc. is a hardware startup that develops mobility products — including the foldable electric bike “TATAMEL BIKE” — built around the concept of a toy-inspired design approach.
・The Reality of AI Adoption: AI has delivered real efficiency gains in visualizing ideas and aggregating information — but ICOMA has also encountered hard limits and missteps, particularly when it comes to qualitative, human-feel output.
・Looking Ahead: ICOMA places “human interaction (Physical AI)” at the center of making AI work in society, and aims to open up Japan’s manufacturing ecosystem.
1. About ICOMA Inc.: Mobility Development That Starts with Toys
ICOMA Inc. is a hardware startup founded in March 2021. Beginning with the development of a toy electric motorcycle, the company went on to develop and sell the TATAMEL BIKE — a foldable electric bike with a distinctive mechanism that transforms from suitcase size into a rideable bike.
ICOMA’s defining feature is its proprietary development approach — making even complex technologies fun and easy to understand through a toy-inspired design approach. Traditional manufacturing tends to involve high prototyping costs and delayed validation. ICOMA flips this by immediately giving every idea physical form as a toy-like prototype, then running rapid, agile cycles of trial and error.
2. AI in Manufacturing: Background and Success Stories
ICOMA has brought AI into its workflow specifically to accelerate this toy-inspired design cycle.
Where AI Has Delivered Results
・Visualizing ideas and raising quality: AI is used as a sounding board when developing ideas, and for generating illustrations for patent documents (such as diagrams showing a person interacting with a robot). The time savings have been dramatic.
・Aggregating and digesting information: When absorbing large volumes of information, using AI to consolidate and summarize reduces cognitive fatigue and makes research significantly more efficient.
3. Learning from AI Failures: Defining the Human-AI Divide
CEO Takamitsu Ikoma also spoke candidly about the failures and obstacles that matter most to readers of AIAM mirAInews working in the field.
Failures and Limits Encountered in Practice
・Handing things off entirely doesn’t work: When trying to delegate emotionally nuanced reporting tasks to AI — feeding large volumes of information into a notebook — the results were wildly off-base and fell far short of expectations.
・The difficulty of qualitative judgment: Processing and evaluating large datasets is AI’s strong suit — but generating output that requires human emotion and sensibility, the truly “qualitative” dimension, remains difficult.
The Lesson from the Field
The answer ICOMA’s team has arrived at: “Going from zero to one, and the final judgment needed to get from ninety to one hundred, both depend on human sensibility.” Deepening your understanding of AI — knowing its structure down to the last bolt — and then crafting precise prompts from that knowledge: that is what separates companies from their competitors.
4. Looking Ahead: Physical AI and “Human Interaction” as the Last Mile
“Physical AI” has become a buzzword across the AI industry. ICOMA defines the last mile of getting AI out into society as human interaction — the moment of physical exchange between a person and a machine.
Consider a simple robot mobility device fitted with wheels. Add just one feature — it responds with delight when you pat it (its expression changes) — and children immediately want to touch it, while the adults around them say, “That’s adorable, I want one.” ICOMA sees its role as that of a translator for society: taking deep tech that most people find impenetrable and rendering it into something lovable — an “irresistible toy (product)” that people are genuinely drawn to.
5. Social Contribution and Talent Development: Building an Open Making Ecosystem
Mr. Ikoma speaks with conviction about his hope that “Japan’s ecosystem for creating new things functions a little better.”
To that end, ICOMA deliberately keeps its development process and technology open rather than guarding them behind closed doors. By making work-in-progress publicly visible, the aim is to grow the number of people — inside and outside the company — who can look at what’s being built and think, “I could make that too.” The ultimate goal is to cultivate a culture of entrepreneurship across Japanese manufacturing as a whole, one where great products emerge naturally.
Editor’s Note (AIAM mirAInews)
“You can’t get something good by handing everything off to AI. In the end, it comes down to human sensibility.” CEO Ikoma’s words will resonate deeply with the many people in manufacturing who have wrestled with generative AI on the job.
ICOMA has found a way to leap lightly over manufacturing’s long-standing challenge — its heavy, slow development processes — by applying their toy-inspired design approach, while deploying AI as a sharp, capable assistant. There are clues throughout this story for how Japan’s making culture might lead the world again. In particular, the insight that “translating technology into lovable interaction is what it takes to bring it to society” seems likely to become an important guiding principle for Physical AI development going forward.