At the Nunatak AI Night in November 2025, a sense of optimism filled the air. Over 100 decision-makers from diverse industries attended not merely to discuss Artificial Intelligence (AI) and Large Language Models (LLMs), but to understand, test, and apply them in practice.The consensus was clear: AI Literacy, understanding essential AI concepts, critically evaluating them, and deploying the technology responsibly, has long been a core leadership priority.

‍

Following the keynote speeches and panel discussion, attendees moved directly into the Nunatak AI Lab. Supported by experts from the Nunatak team, guests explored hands-on applications. How can incoming emails be prioritized and drafted via AI? How can a product video for social media be created in just a few clicks? Which agents can significantly increase efficiency in a call center?

‍

‍

Key Takeaways from the Nunatak AI Night

‍‍

# Getting Real with AI: Executives can no longer afford to avoid testing AI tools and LLMs themselves. They must engage with them regularly because the technology is advancing and becoming reliable at a rapid pace, as Nunatak Founding Partner Robert Jacobi explains. On the one hand, decision-makers serve as role models. On the other, they require a solid understanding of the technology’s possibilities and limitations to make informed decisions. The Nunatak AI Lab offers a safe environment for experimentation, backed by expert support.

‍

# Top-Down, but with a Bottom-Up Approach: A clear AI strategy endorsed by the executive level is essential. However, AI Literacy must also be broadly embedded across the organization, argues Daniel Schiffner (Head of AI at Madsack Media Group). Well-trained, convinced employees act as multipliers and help bring skeptical colleagues along.

‍

# Process Analysis First: Do you have a comprehensive overview of all company workflows? do you know exactly where bottlenecks exist or where data fragmentation causes problems? Anyone answering “No” will struggle to define clear requirements for an AI system. Often, an outside perspective is crucial, explains Thuy-Ngan Trinh (Managing Director at digital agency A11, a spin-off of early-stage investor Project A). It is only through sparring with external partners that vulnerabilities become visible and can subsequently be targeted with AI solutions.

‍

# Guardrails Enable Speed: Governance and clearly defined roles do not slow down AI adoption in fact, they accelerate it, believes Sandy Unbehau (Senior Product Owner Digital & Web Solutions at Rohde & Schwarz). On the contrary, secure guardrails empower teams to apply the technology to high-impact projects. Furthermore, transparent rules help “reduce ‘shadow IT’ (use of unauthorized personal tools or devices) and thus strengthen data security.

‍

# Build vs. Buy – Decide Consciously: Many organizations develop proprietary solutions even when secure, more cost-effective alternatives are available on the market. Follow-up costs are often overlooked in this equation, Gölles points out. In-house solutions require independent maintenance at ever-shorter intervals due to rapid technological cycles.

‍

# An End to “Whataboutism”: “Europe often stands in its own way,” warns Trinh. Instead of leveraging the continent’s strengths, such as in Deep Tech or the defense industry, we get lost in endless debates and skepticism. In dialogue with Jacobi, she appeals to entrepreneurs and policymakers to finally pick up the pace. “It is not yet too late to catch up on missed opportunities and prove our international competitiveness.”