If you lead data or a business unit, here is the takeaway from Tokyo worth your time: start now, and win small.
At the Snowflake World Data Tour Tokyo on September 11, 2026, four companies, Kyocera, Sotetsu Business Service, Trust Base, and Seiko Epson, shared how they are moving beyond traditional dashboards toward agentic AI. The session drew 300 pre-registrations for a room designed to accommodate just 150 people, reflecting growing interest in how enterprise AI can transform everyday business operations.
Each company brought a different frontline challenge and a practical perspective on the evolution of AI and business intelligence. Their experiences point to a common lesson: you don't need perfect conditions to begin adopting agentic AI. You need a focused first step, a measurable win, and a success story worth sharing.
Four Companies, Four Frontlines, One Approach to Agentic AI
Moderated by Toshiharu Makinouchi of Delivery Consulting, the panel moved across four very different operations. What tied them together was not a single use case but a way of getting started.
Each speaker walked through the same three beats: the business challenge they faced, the small step they took with ThoughtSpot, and the agentic AI capabilities they plan to introduce next. Together, their experiences offer a practical look at how organizations can move beyond traditional dashboards toward more intelligent, AI-driven business operations. Here is what each team is doing, and where they plan to take it.
Kyocera: Breaking Silos, Then Adding AI
Kyocera, a Japanese multinational technology and electronics manufacturer, has grown through its well-known Amoeba Management model, where each division operates autonomously. However, over time, data silos emerged, making it increasingly difficult to maintain visibility across divisions. To address this, the company plans to integrate core data in Snowflake and roll out company-wide dashboards through ThoughtSpot, with the goal of giving everyone from executives to the frontline a shared picture of the business.
Today, ThoughtSpot usage centers on dashboard visualization and standardized reporting, and AI features are not yet switched on. Two challenges are pushing the next move. First, data literacy varies widely, so self-sufficient data use stays limited to a small group of staff. Second, dashboard requests from business units keep climbing, and IT resources are stretched thin as a result.
Where does Kyocera go from here? The plan is to bring in Spotter so anyone, whatever their skill level, can ask questions of the data and get trusted answers on their own, with SpotterModel generating the governed semantic model underneath. Alongside that, SpotterViz would proactively surface trends and anomalies, freeing people to focus on interpretation and action instead of hunting through data. Agentic analytics with AI support means the frontline stops waiting in the IT queue, and the queue itself gets shorter.
Seiko Epson: One Semantic Layer on the Warehouse
Kohei Shimada from the P OFFICE & HOME BUSINESS MANAGEMENT DEPARTMENT took on a challenge many data teams will find familiar. Semantic definitions such as metrics, relationships, and business logic were being maintained in duplicate, across both the data warehouse and the BI layer. Every change required edits in multiple places, and users often had to wait through long lead times before newly available data reached them.
Mr. Shimada’s solution was to use Snowflake Semantic Views directly from ThoughtSpot. With that in place, the semantic layer is now managed centrally on the warehouse side. Maintainability improved, and the scope of impact from any model change was reduced. He expects this to shorten the lead time for delivering data.
There is also a secondary benefit. Because Spotter now draws semantic context straight from Snowflake, users get immediate insights from the moment they ask a question. Looking ahead, the roadmap calls for delivering new data and use cases to business-to-business units more quickly and rolling out Spotter while keeping IT operations and maintenance burden low.
Sotetsu Business Service: Consulting for Smarter Bakery Production
Sotetsu Section Chief Shinsuke Horino shares a story you can taste. As part of its ongoing business consulting for the bakery brand "Hayama Bonjour," Sotetsu helped the team adopt ThoughtSpot to visualize sales by product, day, and time of day — giving them insights to adjust production planning based on historical sales, markdown frequency, and sell-through data.
The results already show up on the shop floor. Production volumes are now calibrated more accurately, and markdowns are far easier to track. Where teams once leaned on gut feel, they now base decisions on the numbers, opening concrete steps toward more full-price sales, which is where margin actually lives.
The ambition from here is AI-driven demand forecasting. Picture a model that folds in weather data, local events, day of week, time of day, and markdown history, then recommends the optimal production volume automatically. The payoff Sotetsu is chasing: higher full-price sell-through, fewer stockouts, and a sharper product strategy overall.
Trust Base: Finding the Back-Office Bottlenecks
In financial services, Trust Base pointed the lens inward at its own back-office processes. Takuto Yamamoto described analyzing transaction logs and execution logs to measure processing time, dwell time, errors, and rework across workflows. From this analysis, the team can see which steps are candidates for automation, which still need human judgment, and where the biggest improvement potential lies.
So what comes next? Trust Base wants AI to automate and workflow-ize back-office processes, cutting errors while speeding up root-cause analysis when exceptions crop up.
Yamamoto described the goal as an "intelligence layer": a layer that surfaces anomalies and bottlenecks from trusted data, recommends the next action, and feeds results back into business systems and workflows. Trust sits underneath all of it, since a recommendation is only as good as the data behind it. In a regulated back office, that closed loop from insight to action to system-of-record is exactly where efficiency gains stop being theoretical.
Why This Cross-Industry Lineup Signals Agentic AI's Next Step
Look closely, and a pattern emerges. Each team started with visibility, dashboards, and trusted data, then set its sights on agents that recommend and act: SpotterModel for governed data models at scale, SpotterViz for trusted, instant dashboards, an intelligence layer for back-office decisions, and a warehouse-managed semantic layer feeding Spotter. That is the arc from BI toward agentic AI, told through real operations rather than a roadmap slide.
ThoughtSpot commentator Mashi Yam closed on a bigger idea. His framing was "data monetization": the notion that ThoughtSpot is the platform that helps organizations turn data into value, and a prompt for the audience to ask what their own data is really worth. If your teams are sitting on data they have not yet put to work, that question lands hard, and it is a good one to carry back to your own planning.
For business leaders, the practical benefit is speed and confidence, since decisions from the executive suite to the shop floor now rest on the same trusted numbers. For data leaders, the read is adoption without runaway cost, because a governed semantic layer and self-service agents grow usage while keeping maintenance overhead in check. Both groups saw the same platform, yet each left with a different, concrete reason to act.
Start Now, Win Small: The Shared Takeaway
Four industries, four operating models, and yet one message kept surfacing from the stage: start now, and win small. None of these teams waited for perfect conditions. Each picked a narrow scope, took the first step, and built from there.
Kyocera put the sharpest point on it. Assistant Manager Takayoshi Shibayama argued that the single most important thing is creating "small AI success experiences" across as many frontline teams as possible, then sharing those stories internally. Why does that matter so much? Because the experience of "AI actually worked" lowers the psychological barrier and motivates the next action.
There is a practical design lesson tucked inside that idea. When AI shows up inside the everyday tools people already use, it reads as help rather than one more project to staff and manage. That removes the mental hurdle, and the cycle of broader adoption begins to turn on its own. Stacking small wins and spreading that momentum, rather than chasing one giant rollout, is the surest path to lasting adoption. For a data leader weighing governance, cost, and risk, this is the low-drama route: contained scope, measurable proof, and a story your peers will want to copy.
The clearest signal from Tokyo was a pattern rather than a single product demo. Four teams, four industries, and one repeatable move: start small, prove the win, and let the story carry the next team forward. Agentic AI does not arrive in a single leap, and these frontlines are showing that it compounds, one confident step at a time. Your next step can be just as modest, and just as real.
Ready to see what a first small win could look like on your data? Explore the ThoughtSpot Agentic Analytics Platform and start mapping your own start-now, win-small path today.
