By Masa Maruyama, CEO
From Operations to Intelligence: Where Most AI Initiatives Actually Start
In recent discussions about AI, organizations often begin with models, tools, or new analytical capabilities.
However, when successful initiatives unfold in practice, the starting point often appears much earlier.
Before AI can deliver meaningful outcomes, organizations must ensure that operational data can move reliably across the systems that support daily work.
The Challenge Begins Earlier Than AI
In most enterprises, critical information resides across multiple environments—ERP platforms, operational applications, customer systems, and analytical tools. Each system serves a specific function, but when these environments remain disconnected, it becomes difficult to build a consistent operational view.
Enterprise Systems Were Never Designed to Work Together
Even when dashboards and reports provide visibility, the underlying data may still be fragmented across systems. As a result, teams often struggle to translate information into coordinated action.
Operational decisions rarely live within a single system. They span workflows, departments, and applications.
In many organizations, the challenge is not that data is unavailable, but that the operational systems generating that data were never designed to work together.
When the data flows supporting those decisions are not aligned, organizations may have visibility into their operations without the ability to act on that visibility consistently.
Integration Creates the Operational Data Foundation
This is why many organizations discover that the first step toward AI is not AI itself.
Instead, it begins with establishing a reliable operational data foundation, one that allows information to move across systems, teams, and workflows. Data integration plays a critical role in enabling this movement across operational environments.
When this foundation is in place, several changes begin to occur:
• Operational data becomes accessible across the organization rather than remaining confined within individual systems
• Teams gain greater confidence in the information they rely on for decisions
• New analytical capabilities, including AI-driven insights, can be introduced without creating additional fragmentation
From Operational Data to Real Decisions
We recently saw this pattern in practice through an initiative at Seven Bank.
Rather than beginning with AI experimentation alone, the organization first focused on building a secure data and integration layer connecting internal systems with its analytical environment. By enabling operational data to move reliably across systems, teams were able to introduce natural language analytics capabilities that allow users to explore complex data without relying on technical specialists.
This example reflects a broader pattern we now see across many organizations.
Instead of starting with tools, they begin with operational foundations.
Instead of focusing only on algorithms, they focus on how data moves across the enterprise.
Once that foundation is established, the path toward advanced analytics and AI becomes significantly clearer.
For organizations exploring AI today, the question may not be which AI tool to adopt first.
A more fundamental question comes earlier:
How reliably can operational data move across the systems that support everyday decisions?
For many enterprises, answering that question is where AI initiatives actually begin.
When operational data can move across systems, intelligence becomes actionable. Without that foundation, even the most advanced AI remains disconnected from the work it is meant to improve.
This article builds on earlier perspectives in our From Operations to Intelligence series:
• Visibility Doesn’t Equal Action
• Context Is the Missing Layer
• Why Trust and Governance Enable Action
Together, these perspectives explore how visibility, context, trust, and operational data foundations work together to turn information into action.