Artificial Intelligence is evolving rapidly, but at APAGCoSyst, we believe that speed without a map leads to nowhere. While many are rushing to deploy isolated AI tools, we are taking a deliberate, architected approach, building the robust data foundations required for AI that is reliable, secure, and, above all, impactful.

To go behind the scenes of our digital transformation, we sat down with Johannes Sperber, taking a lead in our cross-functional AI working group. Comprising experts from IT, R&D, Systems Engineering, and Operations, this group is moving beyond the „AI hype“ to build a future where technology doesn’t just automate tasks, it strengthens our competitive edge and elevates the daily work of every employee.

APAGCoSyst is still in the early stages of AI adoption. What are the main goals of the initiative, and what does “building the foundations” mean in concrete terms?

At the moment, our focus is very deliberately on getting the basics right. We are not trying to build highly advanced AI systems or train our own models yet. Instead, we are preparing the company so that AI can actually deliver value when we scale its use.

Many colleagues already use tools like ChatGPT or Claude for everyday tasks such as drafting emails or summarizing text. These tools are helpful, but they operate on general public knowledge and are usually used directly on external platforms, which also raises questions around data security and confidentiality.

For AI to truly support APAGCoSyst’s work, it needs access to structured, company-specific information, be it our documents, our projects, or our production data, in a secure and controlled way. That is the foundation we are building now.

This includes reorganizing engineering and project documentation into SharePoint, enriching it with metadata, and making it searchable and machine-readable. In parallel, we are centralizing production and project data into a high-performance data warehouse. Once these elements are in place, AI systems can deliver accurate, context-aware answers based on APAGCoSyst’s real knowledge, not generic internet content.

It’s also important to distinguish between different types of AI. In this context, we mainly talk about language-based AI systems, such as large language models, which are used for documents, communication, and automation. Separately, APAGCoSyst already applies classic machine-learning techniques in focused areas like automated visual inspection in production. Both are valuable, but they serve different purposes.

You often describe APAGCoSyst’s AI strategy as having three core areas. Can you walk us through them?

We group our work into three main categories to ensure we are covering both immediate needs and long-term infrastructure.

First, text generation and language support.
This is where AI adoption is already well-established. Many employees use these tools for written communication, reporting, and documentation. At this stage, the company’s role is primarily to provide clear guidelines on data protection and responsible usage, ensuring that while we gain efficiency, our sensitive information remains secure.

Second, and our major focus, AI for internal data.
The goal here is to make APAGCoSyst’s internal knowledge usable by AI in a reliable way. Today, data and documents are often distributed across multiple systems. We are addressing this by centralizing production and project data and restructuring technical documentation so it can be searched and linked correctly.

This preparation enables approaches such as Retrieval-Augmented Generation (RAG). Instead of the AI relying on general internet knowledge, it references our specific specifications and reports directly. This significantly improves the accuracy and relevance of the answers for our engineering environment.

Third, AI for tooling and automation.
Several teams already use AI to generate scripts or code. To support this safely, we are building an internal environment that allows even non-programmers to create small automation tools. These tools interact with our systems using standardized interfaces, which ensures they meet our standards for security and maintainability.

Concepts like context engineering, metadata, and knowledge graphs play an important role in your approach. Why are these elements so critical at this stage?

AI is only as good as the context it receives. Without proper structure, even very powerful models can produce incomplete or incorrect results.

Context engineering means deliberately deciding which documents, data sources, and background information are provided to the AI for a specific task. Metadata and structured relationships help the system understand how information is connected, for example, which documents belong to which product, project, or module.

Knowledge graphs extend this idea by explicitly mapping relationships between data, documents, and systems. If someone asks a question about a specific product or production issue, the AI knows where to look because the information is structured and connected.

This approach is the opposite of using AI in an unstructured, experimental way. Our goal is not “trial and error,” but reliable, repeatable results that people can trust in a professional engineering and manufacturing environment.

How will these AI capabilities support employees, and what concrete benefits do you expect for the organization?

One of the biggest benefits will be the reduction of repetitive, low-value tasks. In areas like supply chain or logistics, a lot of time is still spent manually transferring information between emails, documents, and systems. AI can automate many of these steps, allowing teams to focus on what really matters, while preventing issues, solving problems, and coordinating with partners.

Another key benefit is faster access to information. Instead of searching through folders or asking multiple teams, employees will be able to ask AI questions and receive context-specific answers based on APAGCoSyst data.

We already see early examples of this in practice. For instance, one of our brands in the Czech Republic uses an HR chatbot to answer common employee questions through a kiosk interface. This improves accessibility for employees while freeing HR teams to focus on more complex and personal topics.

From a broader perspective, even moderate efficiency gains can have a major impact in an industry with tight margins. For APAGCoSyst, improving efficiency in a controlled way is essential to remaining a reliable, competitive partner. Obrazec

What are the main challenges, and how do you see APAGCoSyst’s AI ecosystem evolving over time?

The biggest challenge is data consistency. Bringing information from different systems into a unified structure is a long-term effort, and it is where most of our current work is focused.

To stay flexible, we are also building a standardized interface layer between our data and AI models. This allows us to adapt to changes in the AI landscape, for example, switching providers without having to redesign all surrounding systems.

Responsible usage is another key aspect. AI should support people, not replace judgment. Clear rules, human review, and employee training are essential to ensure AI is used critically and safely.

Looking ahead, as our data becomes more structured and our systems more connected, we expect to move towards more automated, background processes. These could support quality checks, highlight risks, or assist decision-making in engineering and production. The goal is always the same: reduce indirect effort, improve reliability, and allow our teams to focus on meaningful work.

APAGCoSyst is approaching AI with a clear focus on structure, responsibility, and long-term value. By investing in data foundations, context-aware systems, and internal enablement, the company is preparing for a future in which AI strengthens human expertise rather than replacing it.

This approach not only supports employees internally but also helps ensure APAGCoSyst remains a reliable, competitive, and forward-looking partner in a demanding and fast-changing industry.

Want to learn more about our digital transformation? Sign up for our newsletter and follow us on LinkedIn, Instagram, YouTube for more updates from APAGCoSyst.