Daimler Truck North America operates in one of the most complex corners of manufacturing, designing, and building trucks that move freight across supply chains, along with school buses, chassis for delivery brands, and vertically integrated engines and powertrains. With operations spanning the U.S., Canada, and Mexico, the company sits at the intersection of industrial scale, engineering precision, and operational complexity, making data a critical lever for transformation.

In this first part of a three-part CDO Magazine interview series, Edgar Gallo, Chief Data Officer at Daimler Truck North America, speaks with Susan Wilson of Alation, about how AI agents and metadata are changing manufacturing. The conversation centers on the “how” of transformation: how data leaders build trust, how organizations prepare culturally for agentic AI, and why metadata is becoming the knowledge layer that makes these systems usable at scale.

Redefining the CDO role around influence, not control

As AI moves from experimentation into operational use, Gallo describes a shift in what it means to lead data transformation. In his view, there are two broad models of leadership in the CDO role:

• Built around control: owning the data warehouse, the data lake, the platforms, and the mechanisms needed to orchestrate change from the center.
• Built around influence, curiosity, and collaboration.

Gallo aligns with the second approach. “You have to be one-third humble, one-third curious, and one-third collaborative.”

That mindset, he says, has shaped his leadership approach because transformation does not always begin with formal power. He notes that the CDO role often does not come with direct ownership of every platform, a board seat, or a large budget and team structure. What matters instead is the ability to create momentum across functions and help others solve problems.

For Gallo, that service-oriented perspective has roots in his earlier work in finance and controlling, where progress often depended on exchanging knowledge across domains. A finance leader could help an engineer understand the downstream cost implications of a design decision, just as a data leader can now help the business make sense of the systems and relationships behind its processes. In that model, transformation happens not through command, but through connective leadership.

Building from the catalog toward enterprise knowledge

Gallo explains that Daimler Truck’s transformation began with a practical challenge: understanding what data existed, where it lived, and how it connected across systems. The catalog became an early foundation for that work.

He describes it not simply as a governance tool but as a place of knowledge. Teams needed visibility into what sat behind a business process, what databases supported which applications, and how relationships across environments actually worked. Business users often knew processes by name, while IT and application owners understood only their portion of the stack. Those worlds were not always connected clearly.

The catalog helped bridge that gap. It created a clearer picture of the relationships that existed behind the scenes and gave the organization a better way to understand the data estate it was trying to use. In Gallo’s framing, that work was not only about documentation. It was about making the business legible to itself.

From technical control to outcome-based trust

A major theme in Gallo’s perspective is that AI transformation requires leaders to rethink how trust is built. He argues that data leaders can no longer rely on an old model in which they fully inspect, manipulate, and understand every technical layer before adopting something new.

To illustrate the point, he draws an analogy to vehicle maintenance. In the past, a driver could open the hood, take things apart, and diagnose issues based on a working understanding of the machine. That model has changed. Today’s vehicles are software-driven, diagnostic-based, and updated over the air. The trust model has shifted from mechanical intervention to outcome validation.

“You need to let go of what you thought you could always control and tinker to understand the fundamentals to then go after those new technologies, but evaluate them by the outcome, not by you understanding how they work in the setting.”

That same logic now applies to AI agents. For Gallo, the question is not whether every internal mechanism is visible at all times. The question is whether the system produces accurate, actionable outputs that can support the business with confidence.

Earning trust in agents through rigorous testing

That does not mean trust comes easily. Gallo emphasizes that agentic systems must be challenged before they are accepted. As Daimler Truck North America began testing technologies, the organization deliberately involved people who could push the technology hard. Teams brought in users who understood SQL deeply and could examine, question, and criticize what the system produced. That cycle of testing was essential to establishing confidence.

“We tested within the boundaries of our knowledge and then a little more, and then we said, ‘Done challenging that the engine can write good SQL.’ Now let’s put it to work.”

For Gallo, that is the real threshold. Trust is not blind belief. It is a decision made after enough pressure-testing has taken place to show that the system can perform reliably in practice.

Why metadata becomes the operational trust layer

If trust is the requirement, metadata is the enabler. Gallo argues that in the era of AI agents, metadata matters more because it provides the context needed for interpretation, resolution, and action. Static descriptions of business processes are no longer enough. As agents take on more work, they need a richer and more dynamic understanding of relationships, conditions, and meaning.

“The key here is if we have good metadata, the ontologies are fed.”

In Gallo’s view, strong metadata allows ontologies to become more useful and adaptive. It helps systems understand whether a topic is manufacturing-related, how relationships should be interpreted, and what kind of contextual lens is needed for a given task. The more reliable the metadata, the better the agent can reason within the right frame.

That is why he sees metadata not as a secondary governance artifact but as a core ingredient of AI performance. Poor-quality inputs lead directly to poor-quality outputs.

“Bad data, bad agent. Low-quality data, low-fidelity data. It’s not going to make an informed decision. Metadata is the trade currency, not only for our trust but also between agents and their operational trust.”

Culture, not technology, becomes the turning point

While AI and metadata play a vital role, Gallo also identifies culture as the real inflection point in Daimler Truck’s journey. He points to an earlier initiative called Data Days, which helped create an internal community around data long before agentic AI entered the picture. At the time, many organizations wanted to become data-driven, but for Gallo, the more important outcome was the subculture that formed around people who were genuinely interested in data, databases, and more sophisticated ways of solving business problems.

Over time, he saw a meaningful shift in organizational behavior. Instead of simply asking IT whom to contact for access to data, employees began asking deeper and more technically informed questions about tables, queries, and how to retrieve the right information themselves. That was a signal that the data culture was maturing.

Concluding, Gallo highlights the growth of an employee resource group, which began in a small part of the business and has since expanded significantly with 900 members now. What started as a focused effort among people trying to improve fulfillment and business processes through data has grown into a much broader internal network of data-curious employees.

CDO Magazine appreciates Edgar Gallo for sharing his insights with our global community.