Data is the fuel of modern manufacturing, driving innovation, efficiency, and competitive advantage. As data becomes more complex, manufacturers face acritical question: Am I drowning in data or swimming in insights?

Joost van den Dool, Director of Architecture Global Connected Services at PACCAR, shares his expertise on the challenges and opportunities of data and AI in manufacturing, together with real-life use cases.

Want more insights? Watch the virtual insights session, Data Transformation in Manufacturing: Powering Industry 4.0 with AI.

Hi Joost! Tell us more about PACCAR and your role

PACCAR is an international holding company of truck manufacturers based in the U.S. It operates three major truck brands: Kenworth, Peterbilt, and DAF. Kenworth and Peterbilt focus on the American market, while DAF targets the European market. I work in the Global Connected Services division, where we collect vehicle data and process it into insights for our customers. I lead a team of architects globally, with members in Europe and North America. Together, we design solutions to transform vehicle data into actionable insights, ensuring our customers have the data they need to succeed.

Is AI driving decision-making or vice versa?

I believe that AI can facilitate decision-making but should not make the decisions . While AI tools are advancing rapidly, I prefer maintaining control. For example, in vehicles, we already use tools, whether rule-based or AI-driven, to assist drivers in safer and more efficient driving. Ultimately, the driver remains in control. They are responsible for ensuring safety, but AI and data can provide valuable support.

What is the relationship between AI and data?

AI cannot function without quality data. First, you need to collect reliable vehicle data and bring it into a backend system. Our initial use cases often involve providing insights, such as understanding vehicle performance or assessing its health. With quality data, you can go further. For instance, AI-driven solutions can assist drivers with lane departure warnings or alert them to nearby vehicles. On a larger scale, algorithms can predict component wear and schedule maintenance proactively, maximizing vehicle uptime.

What competencies are needed to keep decision-making within the human domain while leveraging AI as a tool?

Judging AI algorithm outcomes is a crucial skill. AI can reflect biases based on the data and algorithms used. It’s vital not to accept results blindly but to validate them with human knowledge, beliefs, and experience to make the right decisions.

How can we use AI to leverage growing amounts of data and keep swimming in opportunities rather than drowning in challenges?

In some cases, we’re successfully leveraging data to provide value, while in others, significant untapped potential remains. A key step is organizing data effectively. By using a data lake, we can filter and prepare the necessary data for AI processing. However, data management is complex—it starts with collecting data from a vehicle’s electronic unit and continues through multiple stages. If data is missing, we may need to revisit the source. This process requires ongoing adjustments to balance business needs with data availability efficiently.

Can you provide a practical example of transitioning from feeling “drowned” in data to finding opportunities through data utilization?

For instance, when faced with an issue, we analyzed data from the lake to help engineers investigate and pinpoint the root cause. Similarly, as cybersecurity legislation becomes stricter, we’ve adapted to provide relevant vehicle data while adhering to compliance requirements. By identifying and loading only the necessary data, we ensure our trucks meet safety standards.

What makes a successful data governance program when it comes to privacy and compliance?

Do it together. If one person in the organization starts to be an enthusiast about data governance, it will fail. That’s a difficult part of data governance. You need to find momentum as an organization to take that step. External motivators, such as GDPR, often drive awareness and action. At PACCAR, internal motivators like data quality also play a role. As a company committed to reliability, we recognize the need for high-quality, well-governed data. My team includes data architects who bridge the gap between the company and customers, ensuring continuous improvement in data governance practices.

How do you ensure security around AI solutions to prevent data tampering or manipulation?

PACCAR is an organization that prioritizes security—it’s embedded in our DNA— and this begins with the vehicle itself. Protecting the vehicle from tampering is essential, both to ensure safety and to comply with increasingly stringent laws. To address this, we implement measures to isolate critical components of the vehicle and make them as secure as possible.

Data encryption is used during transmission, along with non-repudiation algorithms to ensure data integrity. Once the data reaches our back office, it undergoes validation and verification to confirm its reliability. In countries where data collection is restricted, we ensure that our central environment processes and cleanses the data to meet security and compliance requirements.

When applying AI algorithms to this data, whether in-house or through a partner, we enforce robust security measures to ensure that the data remains safe and compliant at all stages of use.

What is the role of data and analytics in enhancing customer experience and forecasting future trends at PACCAR?

Our slogan, “Simplifying your life,” reflects our commitment to supporting both drivers and fleet owners. For drivers, we use data to optimize route planning, factoring in vehicle range and driving comfort. Data-driven tools help drivers navigate safely, monitor their surroundings, and improve efficiency—saving fuel and enhancing safety. For fleet owners, our data solutions maximize vehicle uptime, ensuring trucks are operational and reducing downtime.

Can you provide a specific example of how collecting data led to an improvement or outcome that impressed you?\

The Eco Scorecard is one example of how data improves fuel efficiency. Initially, we feared drivers might view it as intrusive, but they appreciated the coaching it provided. This interaction extends to fleet reports, which deliver actionable insights. Predictive maintenance is another growing area, further improving vehicle uptime.

What is the role of data and analytics in building a solid business continuity program?

While data serves as an enabler and provider of insights, it also poses challenges for business continuity. If an incident occurs and systems need to be restored, how do you ensure that all distributed components maintain a consistent view of the data and remain synchronized?

At the same time, our reliance on data has grown significantly, making its availability even more critical. The increasing use of cloud solutions helps address part of the problem by improving uptime and simplifying efforts to enhance availability. However, organizations must still prepare for potential incidents and ensure rapid recovery.

This presents a new set of challenges, particularly with distributed components. Systems need to be designed with resilience in mind to ensure that at least customer-facing operations remain available as quickly as possible, followed by the restoration of internal structures as necessary.

Can you provide a practical example of how using AI or data led to a significant improvement in a process, product, or service for your customers?

Let’s talk about navigation. In navigation, we adapt route information to accommodate the unique requirements of trucks, such as road restrictions. Integrating navigation systems with fleet back-office systems allows for dynamic updates, like additional stops or optimizing routes for electric vehicle ranges or hazardous goods. These efforts enhance efficiency and safety for our customers.

How do you evaluate both internal and external factors that could impact decision-making when using data?

As a company, we are accustomed to accounting for risk, especially since we build trucks, and safety is always a priority. We work extensively with Six Sigma, which includes a valuable tool called FMEA—Failure Mode and Effect Analysis. This tool is particularly useful when dealing with high-risk situations. It allows us to thoroughly examine the solution or process to identify potential failures, whether from internal or external factors.

By using FMEA early in the design phase, we can proactively address potential issues and incorporate solutions, ensuring risk management is integral to the process rather than an afterthought. This approach works best when applied across different scenarios and involves people from diverse backgrounds, enabling collaboration from multiple perspectives to address challenges comprehensively.

Is communicating your data-driven insights and processes to nontechnical stakeholders within your organization and customers a challenge?

Yes, it’s a challenge, but I also see it as an opportunity. I’m trained in the Scaled Agile Framework, which brings the business representative, the project manager or team implementing the solution, and the architect together. I believe this is highly effective because while the role of an architect is essential, architects can sometimes go too deep into details. Combining these different disciplines ensures the best balance of interests: the architect focuses on creating a future-proof solution, the team ensures it’s practical and realizable, and the business representative ensures their needs are met. By working together, you can turn challenges into opportunities and achieve the most balanced and effective results.

What is your recommendation if the data quality is poor – focus resources on fixing the data sources or look for alternative data sources?

It’s a difficult question. Balancing immediate needs with long-term goals is a common challenge. As an architect, I advocate solving problems at their source for sustainable benefits. While workarounds may be necessary in the short term, addressing root causes ultimately reduces complexity and technical debt.

What are the key low-hanging fruit implementation areas in the context of Industry 4.0 and AI?

My first suggestion is to ensure you have data. It doesn’t need to be a lot, but having some data, preferably of good quality, is a strong starting point. In this context, taking an agile, step-by-step approach is key. Start small and create prototypes. From an architecture perspective, there are now solutions you can build upon as you grow. This allows you to learn, gain insights, and market the concept within your organization, helping to build momentum. From there, you can expand gradually, which is the most important aspect.

Data always feels abstract, so you need concrete examples and successes to sell it to the organization. AI operates on different levels. Start small, perhaps with a basic algorithm, before progressing to machine learning. This helps you understand your data, take incremental steps, and involve your customers in the process.

What are the key developments in data and AI that you think we should watch for?

We’re only at the beginning of what data and AI can achieve, especially when considering tools like ChatGPT. It marks the start of a new era. There’s tremendous potential for automation and innovation using data, but this also presents risks. With access to data, individuals with harmful intentions can misuse it and gain insights for negative purposes.

How do we differentiate between good and bad actors? How do we ensure data is interpreted correctly? These are key challenges. The potential is incredible— we already see its impact in small ways, like Spotify recommendations, and larger applications, like driver-assistance systems in vehicles. But we must continuously evaluate the balance: is the human or the machine in control?

Ultimately, computers rely on data from sensors; they cannot think or interpret the way humans do. As a society, we must learn how to work with these technologies responsibly and develop ways to manage their use effectively.

*The interview answers have been edited for length and clarity.