grubel
Automated AI adaptation for complex legal matters.
The founders
RH
MH
The company
Grubel is an AI research lab building matter-specific AI for complex legal work. Its system combines a data engine, test-time adaptation to each matter and continual evaluation, so clients can run confidential, case-specific models on their own infrastructure.
In September 2026 Grubel raised a €3 million pre-seed round led by Point Nine, with angels including Jeff Dean, Chris Ré, Ion Stoica and co-founders of Harvey. Running models on the client's own infrastructure addresses the confidentiality that legal work requires.
Grubel was founded in 2026 as a Munich and Tübingen research lab by Moritz Hardt and Reinhard Heckel.
Moritz Hardt is at the Max Planck Institute for Intelligent Systems, was previously at UC Berkeley and Google Brain, and studied at Princeton University. Reinhard Heckel is a professor of machine learning at TUM and previously worked at IBM.
Who else is building in this space
Companies working on the same thing, at every stage.
Unicorn
Agentic AI workspace automating legal review and drafting. | Formerly known as Leya
Unicorn
Autonomous AI agent executing complex tasks with minimal human input
Unicorn
Tokyo-based AI lab building orchestration models that command a pool of LLMs
Unicorn
Formerly known as Metaphor Systems. | AI-native search engine for applications and workflows.
Unicorn
Real-time APIs for speech-to-text, text-to-speech, and voice agents
Scale-up
Enterprise conversational AI for customer service calls
Thousands more founded every month.
Dealroom tracks new companies as they appear — founders, funding and signals, over the API or in the platform.