VDI Conference "AI in Production" 2026 – A Recap

This year's VDI expert conference on "AI in Production" has come to a close – and, as in previous years, it was well worth attending. For the third time in a row, we experienced a spirit of momentum reminiscent of the inaugural conference in Augsburg in 2024. The reason: agentic AI in production.

From Tool to Active Participant

A central theme running through the first day of the conference: AI is evolving from a passive tool into an autonomous actor. AI agent systems can plan, execute, and deliver results independently – a fundamental shift in the role of AI within business processes.

Multi-agent architectures have emerged as the key paradigm. At the center sits an orchestrator that coordinates specialized agents – memory, reasoning, perceiving, and learning agents. Graph-based structures define allowed interactions between agents, enabling controlled, scalable collaboration.

The practical applications presented were diverse and immediately relevant: from automated quote generation to processing heterogeneous inputs from email, PDF, and handwriting, to supply chain risk management with risk-assessment and action-recommendation agents, and AI-supported quality inspection.

The Data Foundation as the Decisive Success Factor

What all these use cases have in common is that they depend on a robust data infrastructure. And that was precisely the recurring theme of the conference – one that points beyond technical implementation.

The materialization of idiosyncratic knowledge – that knowledge inherent to a company and its processes – into data form is no minor task (cf. Bunk). It is equivalent to potential innovation leadership and a strategic investment in the future – for companies as much as for research institutions.

Company-specific know-how, embedded in processes, decision rules, and the experiential knowledge of employees, is exactly what European companies must leverage to withstand international competitive pressure, compensate for demographic shifts, and maintain their values.

Strengthening Infrastructure and Collaboration

Indispensable during this transformation is building infrastructure to connect data across the organization. Isolated information silos – local accumulations without connection to each other – cannot be effective in automated processes. Only networking makes data usable.

Multiple lectures also emphasized that collaboration between domains must be strengthened: process understanding and programming expertise need to work more closely together. Neither perspective alone is sufficient.

Tooling is Maturing – the Organization Must Follow

Tooling ecosystems are maturing. Solutions from both large and small players now enable organization-specific agent creation. The consensus among experts was clear: data foundations and organizational readiness precede technical implementation. You can build the most advanced agent architectures – without trustworthy, well-structured data as a foundation, they remain empty.

As knowledge sources, anything the imagination allows comes into question: employee experience, work instructions, maintenance and inspection reports, operation logs, video and image recordings, sensor data, expert interviews – and much more. The associated tasks are demanding: structuring, versioning, consistency management, context provision, annotation, and evaluation.

Outlook

We are eager to see how industry – and particularly the Mittelstand (mid-sized businesses) – will drive this transformation forward. And what topics we will be discussing at next year's expert conference on AI in Production.

One question lingers: Will agents in the future not be guided by humans, but instead independently shape entire workflows and "hire" human resources only when needed?

How We Can Help Along This Journey

The VDI conference made one thing clear: the success of AI transformation depends on both technology and foundations – data quality, process understanding, and the connection of domain expertise with technical implementation.

This is precisely where we operate. We help companies build their data-driven infrastructure, translate idiosyncratic knowledge into actionable data, and develop AI solutions that fit into existing processes.