Munich and Tübingen based artificial intelligence research lab Grubel has raised €3 million in pre-seed funding to advance systems that adapt to individual legal matters. The financing was led by Point Nine and attracted a group of prominent angel investors with expertise in artificial intelligence and legal technology. The fresh capital will support research, product development, and team expansion.
Founders with Deep Research Background
Grubel was founded in 2026 by machine learning researchers Moritz Hardt and Reinhard Heckel, who bring strong academic and industrial expertise to the startup. Hardt is a director at the Max Planck Institute for Intelligent Systems and a former UC Berkeley professor and Google Brain researcher. Heckel is a professor of machine learning at the Technical University of Munich, currently on leave and formerly a researcher at IBM Research.
Investor Backing and Market Validation
Point Nine, an early-stage investor active in the Berlin startup ecosystem, led the round. The financing attracted a prominent set of angel investors, including Jeff Dean, Chris Ré, Ion Stoica, and Harvey co-founders Gabe Pereyra and Winston Weinberg. Point Nine partner Louis Coppey said the team is pushing boundaries in legal AI and could unlock new capabilities in complex knowledge work over time.
Why Legal AI Has Lagged
According to Grubel, complex knowledge work beyond coding still relies heavily on highly skilled human labour, with artificial intelligence playing only a limited role in legal services. The company argues that legal AI has lagged because complex legal work is highly specialised and difficult to evaluate, often requiring expertise tied to a specific matter's facts, documents, terminology, and client standards. Building a separate AI system for every matter is resource intensive and not sustainable for engineering teams.
Automating the Specialisation Process
The company's central thesis is that every legal matter needs its own AI, and it is automating the process of specialising an AI system for each case. Its specialisation loop consists of a data engine that curates relevant information, a test-time adaptation layer that adjusts the model and agent before work begins, and a continual evaluation framework based on matter-specific standards. The loop repeats until the system's output meets the required legal standard, reducing the need for manual customisation.
Early Results and Capital Allocation
An initial evaluation reported by Grubel showed that its specialisation loop can improve performance over general-purpose frontier systems on the legal tasks tested. The company plans to use the new funding to further develop the automated specialisation loop for complex legal work while enabling more routine tasks to be handled by smaller, more efficient models. The investment will also support continued research, product development, and growth of the team.
This pre-seed round stands out for its combination of a focused technical thesis and validation from investors with direct experience in AI infrastructure, research, and legal applications. Grubel is targeting a genuine structural bottleneck in applying general-purpose models to matter-specific professional services, where context and evaluation standards vary from case to case. If the approach continues to show strong results, it could have implications beyond legal work for other complex knowledge domains.