This German AI Startup Just Raised $13.1M to Build AI That Never Needs Retraining

Jan de Vries · 2026-07-23

### A New Kind of AI: One That Learns Like You Do Most AI models today are like students who cram for an exam and then freeze when the questions change. They ingest massive datasets, train for weeks, and hope the world doesn't shift too much before the next update. But what if AI could adapt on the fly, just like humans do? That's exactly what kausable, a German DeepTech AI startup, is building. The company just raised $13.1 million in a Seed round to develop a "reasoning-first" frontier AI that adapts to changing contexts without needing to be retrained. Think of it as an AI that doesn't just memorize the past but actually understands how to reason through new situations. ### Who's Behind the Money? The round was led by UVC Partners and Entourage, with follow-on investments from German investors HTGF and Mätch VC. A group of private angel investors from the AI world also chipped in, including folks from Black Forest Labs, OpenAI, Google DeepMind, Noxtua, and the European Laboratory for Learning and Intelligent Systems (ELLIS). That's a serious vote of confidence from some of the brightest minds in the field. ### The Problem With Today's AI Let's be real: current AI systems are fragile. They're trained on static datasets and retrained every few months (or weeks) to keep up. But the world doesn't move in neat, predictable cycles. A sudden market crash, a new virus, or a supply chain disruption can throw everything off. Traditional AI just can't handle that kind of curveball. Johannes Haux, CEO and co-founder of kausable, puts it this way: "The world evolves rapidly, and so should AI. However, even the most capable current AI systems are very static in how they interpret the world and need constant, time-consuming, and costly retraining. With kausable we solve this problem by developing a new kind of world model that learns efficiently and robustly on its own and adapts to the ever-changing world." ### How kausable's AI Is Different kausable was founded in 2025 by three physicists with ties to Heidelberg University and Black Forest Labs: Johannes Haux (CEO), Dr. Benjamin Herdeanu (CTO), and Gregor Ramien (COO). They're building what they call a "world model" — a foundational AI that learns from minimal data and adapts to scenarios that weren't even imagined when it was built. Here's what makes it special: - **No retraining needed:** The model updates itself as conditions change. - **Learns from minimal data:** It doesn't need millions of examples to get smarter. - **Handles the unexpected:** It can reason through problems it has never seen before. The company recently developed TipPFN, a forecasting model that can predict "black swan" events — rare but highly impactful occurrences in fields like medicine and energy. They also raised a pre-Seed round of $1.6 million to get started. ### A Human-Like Approach to Learning kausable's reasoning-first AI learns similarly to humans. You know how you can walk into a new kitchen and still figure out how to make coffee, even if the layout is different? That's because you have a mental "world model" — a set of intuitions that help you adapt quickly. kausable's AI works the same way: it builds a robust, universal understanding of how things work, then applies that understanding to new situations with very little new information. ### Validation From Top Researchers The team recently co-authored a research paper with experts from Columbia University that validates the causal reasoning architecture behind their model. That's a big deal. It means the science checks out, not just the hype. Pieterjan Bouten, co-founder of Entourage, sums it up well: "Most AI models are trained to remember the past. kausable is building AI that can reason about the future. Instead of relying on ever-larger datasets and constant retraining, they're developing a fundamentally different approach: systems that adapt, infer causality and solve problems they have never seen before." ### What's Next? With the fresh funding, kausable plans to expand its current nine-person team of experts and accelerate development of its rapid-learning frontier model. If they succeed, this could change how we think about AI — moving from brittle, single-purpose tools to systems that genuinely understand and adapt to a changing world. For industries like robotics, energy, and finance, that's not just an upgrade. It's a whole new way of thinking.