AI Cologne

If you’re running a Large Language Model (LLM) in production, you know the truth: AI safety is not an optional feature, a nice tool to have, something you can negotiate. Clearly, it's a key design consideration, an architectural layer we must build in from the beginning. Relying solely on the base model's pre-training is naive at best, a strategy for disaster in instances. We need clear, technical, and layered defenses. This article breaks down the inherent risks we must address when designing LLM-centric applications and systems. We provide an illustrative example on how to concretely detect and mitigate a threat using an open model. While we’ll illustrate this using the Gemma 3 family of models, the principles and tooling apply across the board—remember, other powerful open-source guardrail models like Llama Guard and Qwen3Guard exist. Qwen3Guard is a recently published and highly capable model. 1. Understanding the Fundamental LLM Risk Modes LLMs are inherently p...

In an age of information overload, many of us are turning to platforms like YouTube for quick tutorials, in-depth explanations, and even culinary inspiration. Need to master a new software? Fix a leaky faucet? Or perhaps bake the perfect batch of cupcakes for a weekend treat? A quick search on YouTube and you're presented with a plethora of video guides. However, this convenience comes with a trade-off. The trend of creating video content, or "vlogging," has, in some cases, replaced the art of the well-written blog post. We've all encountered them: lengthy videos that painstakingly detail simple instructions, sometimes with nothing more than a creator typing into a Notepad window. What could have been a concise, skimmable article is now a three-hour video commitment. This is where the power of artificial intelligence, specifically multi-modal generative AI, comes into play. We are already witnessing the prowess of Large Language Models (LLMs) in summarizing extensive ...

The field of technology is characterized by ongoing development, and the current advancements in Artificial Intelligence (AI) are bringing about notable changes. For technical professionals, including data scientists, simulation specialists, and software engineers, staying informed about these trends is essential not only for personal and career development, but also to anticipate and creating effective solutions to existing and more importantly novel problems. Recently, I came across a great talk of Bret Taylor co-founder of Sierra (been with key role at Google, CTO of FB), called " How AI is reinventing Software Business Models " - a clear recommendation. In this video Bret talks about Key trends and what this implies for how enterprises engage with customers, the implications for software and the crafts- and salesmen thereof. Below, I will briefly discuss these. Key Developments in the Application of AI The way AI is being integrated into business processes and product of...