7.10.2025 – 4. Workshop Future Skills & AI

On 7 October, the 4th webinar on Future Skills & AI took place in the thematic focus area Work 5.0, which is organized by Platform Industry 4.0 Austria as part of the Interreg AT-SK project “Twin City Future Innovation Manufacturing Hub”.

Roman Kern, Chief Scientific Officer at Know Center Research GmbH and Assoc. Prof. at Graz University of Technology, gave the first presentation on Agentic AI and provided insights into his research area. Modern AI technologies such as LLMs, generative AI and similar technologies such as visual transformers can already be used well, they understand language, analyze images and can draw simple conclusions, generate images and texts themselves. This opens up new application possibilities in search, document understanding and knowledge management and can increase quality and productivity. Despite the progress, challenges remain: AI invents facts (hallucinations), knowledge gaps arise with outdated training data. There are systematic distortions and errors with bias, or a question is always answered differently.

Agentic AI describes AI systems that independently perform tasks, make decisions and control processes – with the aim of supporting people and increasing productivity. In contrast to agentic AI, AI agents today primarily support systems that automate workflows and work in the background. One practical example of AI is support for business travel processes. AI automatically checks boarding passes or receipts and increases efficiency without fully automating.

The vision of agentic AI, where several agents solve complex tasks autonomously, is not yet fully possible in practice and is the subject of research. In a new research project on AI, a comet module, which will start in 2026, the Know Center will continue its research on Agentic AI and, for example, look at how AI can interact with each other and with humans. Today, agentic AI is supportive, not autonomous. Humans remain indispensable and responsible use is crucial.

The second input from Matthias Hayek, Group Leader for Assembly Planning and Assistance Systems at Fraunhofer Austria Research, was dedicated to the AI-Mobile. This project brings artificial intelligence to life in a practical way, demonstrates specific potential applications and strengthens employees’ skills in dealing with new technologies. The AI-Mobile brings AI directly to the Austrian process and manufacturing industry. As a mobile innovation platform, it brings demonstrators and workshops directly to companies, where real AI applications are presented, for example in the areas of quality control, production planning or purchasing optimization, and new ideas are developed together. The aim of the project is to make AI understandable and tangible, to demonstrate practical fields of application and to support companies with technical, organizational and legal issues. Practical examples range from AI-supported quality control with minimal training requirements, chatbots for company data and sensor-based production planning to language models for safety and training applications. The focus will also be on future skills such as prompt engineering, human-in-the-loop and agile project management for AI processes.

Registration and further information on the AI-Mobile can be found HERE

The final input was given by Daniela Rodriguez Ulloa, Enterprise Solutions Manager at Hutchison Drei Austria, who presented the development of the first AI-supported B2B voice product. The background to the project is the new EU AI Act, which fundamentally changes the requirements for companies. In future, AI expertise must be firmly anchored in all areas of the company, from technology to legal & compliance. The aim was therefore to develop a voice system that can be operated independently and developed flexibly. The project not only strengthened technological expertise at Drei, but also promoted cross-divisional and agile collaboration, with a clear focus on continuous learning. The most important findings were: AI affects all teams, which is why a company-wide understanding is necessary; compliance by design ensures that data protection and transparency are considered from the outset; and agility and a learning culture are crucial, as successful AI projects thrive on experimentation and interdisciplinary collaboration. In the future, it will be crucial that technology, business and legal teams work even more closely together. After all, those who are willing to learn and take responsibility will make the difference in the age of AI.

12.11.2025 – 5. Workshop Future Skills & AI

On 12 November, the 5th webinar on Future Skills & AI took place in the thematic focus area Work 5.0, which is organized by the Industry 4.0 platform as part of the Interreg AT-SK project “Twin City Future Innovation Manufacturing Hub”.

The speakers were:

• Benjamin Schwärzler, CEO, Workheld GmbH
• Philipp Bousa, Vice President CIO & Digital Office, OMV Group
• Markus Wagner, St. Pölten UAS, Head of the Josef Ressel Center – Industrial Data Lab

Input Benjamin Schwärzler, CEO Workheld GmbH – Use Case: Swiss Post

Workheld, a Viennese software company with a focus on production and maintenance processes, has worked with Swiss Post to implement an AI-supported use case for the sustainable safeguarding of specialist knowledge. The starting point was the key challenge that experienced maintenance experts for the sorting systems will be retiring in the coming years, but their expertise is essential for the continuous 24/7 operation of the systems.

In collaboration with Swiss Post, an AI solution was developed that digitally maps this company-specific expert knowledge and supports around 300 maintenance technicians at all times. Existing data sources such as fault logs from the expert hotline, SAP data, historical maintenance orders and manufacturer documentation were processed in a structured manner, linked together and merged into a knowledge graph.

The result is digital second-level support that secures critical know-how in the long term, supports technicians directly in the work process and ensures Swiss Post’s high system availability in the future. Another aspect is the multilingualism of the solution: In the final version, the AI will be available in German, French and Italian. This takes account of Switzerland’s trilingualism and makes documentation and specialist knowledge accessible to all employees, regardless of their language of origin.

Input Philipp Bousa, Vice President CIO & Digital Office, OMV Group

The presentation provided insights into the structured development and use of GenAI at OMV over the past two and a half years. The aim of the approach was to initially develop broadly applicable and standardized AI use cases and to gradually develop these into highly specialized, effective applications.

The basis for this is a clearly defined governance and framework approach with a focus on responsible AI, information security, data governance, technology partnerships and platform strategy. Initial use cases such as an in-house GPT or sentiment analyses for multilingual evaluation of IT and employee feedback were quickly implemented and successfully scaled to other areas such as HR or retail.

At the same time, OMV made targeted investments in empowerment: training courses, workshops, prompting training and show & tell formats enabled the specialist departments to develop their own AI ideas and use GenAI sensibly in their day-to-day work.

Today, OMV works with reusable GenAI building blocks – such as for translation, document analysis, RAG or summarization – which are provided via a central AI hub. This modular approach makes it possible to implement increasingly complex use cases through to agentic AI solutions and thus create sustainable added value for the entire company.

Input Markus Wagner, Head of the Josef Ressel Center – Industrial Data LabSt. Pölten UAS

The Josef Ressel Center “Knowledge Visual Analytics for Industrial Manufacturing Data” develops solutions to preserve expert knowledge in industry, use large amounts of data efficiently and reduce the cognitive overload of employees. The aim is to optimize production processes, shorten set-up times and support quality assurance.

The approach is based on visual analysis methods (knowledge-assisted visual analytics): Expert knowledge is digitized, transferred to knowledge graphs and combined with causal models so that problems in machines or processes can be analysed in a targeted manner. A central component is the human-in-the-loop, which validates and feeds back knowledge in order to avoid errors in the model. The knowledge gained from this should make it possible to optimize planning and production processes and achieve better product quality.

Three central use cases were presented:

1. Optimization of machine settings
2. Sales and Operations Planning
3. Root cause analysis (error cause detection)

In addition, prototypes for visual analytics in production, material flow and logistics are being developed that enable AI-supported analyses, chatbot interaction and prioritization according to economic relevance. The aim is to make industrial processes more efficient, secure knowledge and make the solutions available as guidelines across all sectors.