Kausable Raises €12 Million Seed Round to Build AI That Adapts Without Retraining
Updated
Updated · EU-Startups · Jul 23
Kausable Raises €12 Million Seed Round to Build AI That Adapts Without Retraining
3 articles · Updated · EU-Startups · Jul 23
Summary
€12 million in seed funding will let German startup kausable expand its nine-person team and push ahead with a reasoning-first frontier AI model.
The company says its core technology is a “world model” that learns from minimal data, adapts to changing conditions on its own and avoids the frequent retraining used by many current AI systems.
UVC Partners and Entourage led the round, with follow-on backing from HTGF and Mätch VC plus angels tied to OpenAI, Google DeepMind, Black Forest Labs and ELLIS.
Founded in 2025 by three physicists linked to Heidelberg University and Black Forest Labs, kausable previously raised €1.5 million pre-seed and recently published research with Columbia University on its causal reasoning architecture.
Kausable is targeting dynamic fields such as robotics, energy and finance, and points to its TipPFN forecasting model as an example of handling rare, high-impact “black swan” events.