On Thursday, 18 September, TrustLLM joined the very first Jülich LLM Day at the Jülich Supercomputing Centre (JSC). Co-hosted with TrustLLM and HPC Gateway, the event gathered experts from academia and industry to discuss the opportunities and risks of deploying LLMs responsibly in a European context.
The day opened with an informal networking lunch. More than 50 participants from 40 organisations connected with peers and exchanged perspectives on artificial intelligence in business and research. The relaxed atmosphere allowed for conversations that would later shape the discussions in talks and panels.
Keynote: Trustworthy and European LLMs
The official program began with a welcome by Stefan Kesselheim, Head of JSC’s Simulation and Data Lab Applied Machine Learning. He emphasised the strategic importance of trustworthy AI in Europe, where data privacy, transparency, and security are central to innovation.
Following the welcome, Jan Ebert delivered a keynote titled “Trustworthy and European LLMs in TrustLLM.” He outlined the risks of relying heavily on models developed outside of the EU. These risks include potential violations of European data protection standards, hidden security threats, and a lack of alignment with local market needs. Ebert also described the technical and organisational hurdles in building competitive European alternatives, stressing that sovereignty in AI development is critical for long-term innovation and independence.
Jan Ebert, Jülich Forschungszentrum
JSC’s Role in Building Bridges
Next, Sohel Herff, Fritz Niesel, and Maria Petrova-El Sayed presented JSC’s ongoing efforts to strengthen industry collaborations. They highlighted partnerships enabled through initiatives such as WestAI and JAIF, which connect researchers with industry to tackle practical AI challenges.
Oliver Bücker introduced the MATSE apprenticeship program, an initiative designed to equip young professionals with the skills to shape the next generation of AI systems.
Beyond Buzzwords: LLMs in Action
The middle of the day shifted from vision to reality. Three companies pulled back the curtain on how they’re actually using LLMs:
- Philip Reissel (amberSearch) shared his company’s experience integrating LLMs into enterprise environments. He described both the benefits and the hurdles – such as aligning LLM output with company-specific knowledge and ensuring accuracy in sensitive domains.
- Clemens Hasler (u-form) offered a compelling transformation story. His family-owned printing business successfully developed AI-driven exam preparation tools for apprentices. This transition highlighted how even traditional industries can embrace LLMs to stay competitive and relevant.
- Omid Hashemi (kumo) expanded the conversation beyond chatbots. He described LLMs as “process partners” – tools that can optimise workflows, support decision-making, and drive innovation. Hashemi emphasised the importance of deeper integration, arguing that AI’s real value emerges when it becomes part of a company’s operational fabric.
An Open Conversation
The afternoon concluded with an interactive panel featuring Jaroslaw Kutylowski (DeepL Founder & CEO), Valentina Kerst (Senior Managing Partner, KI Bundesverband), and Stefan Kesselheim (Head of JSC’s Applied AI Lab).
What made the panel unique was its open chair format. Two additional seats were reserved for attendees, who could step in and contribute to the discussion at any time. This setup fostered a dynamic dialogue, breaking down the barrier between speakers and audience. Topics ranged from Europe’s competitive position in the global AI race to regulatory frameworks and the need for more transparent and explainable systems.
Closing & Outlook
The event wrapped up with informal networking over drinks, allowing participants to reflect on the day’s discussions and connect with new partners. A guided tour of JSC’s supercomputing facilities offered a glimpse into the infrastructure powering many of Europe’s most ambitious AI projects.
Jülich LLM Day showed what happens when scientists, entrepreneurs, and policymakers share a room: honest debate and fresh ideas for the path ahead.
The first edition set the bar high. If Europe is serious about shaping AI on its own terms, gatherings like this aren’t optional – they’re essential.