Paul Permantier

Paul Permantier is co-founder and CEO of LexGraph. His goal is to enable AI systems to answer legal questions correctly and with reliable sources. To this end, he and his team are developing an infrastructure that structures legal data, links it together and makes it usable for AI applications.

He studied computer science, mathematics and physics at Humboldt-Universität zu Berlin. During his studies, he focused in particular on machine learning and information retrieval and worked as a web developer and freelance programmer. As early as 2023, he developed an application that used large language models to help students summarise video lectures.

The starting point for LexGraph was his bachelor's thesis analysing German court decisions using Retrieval-Augmented Generation. In it, he examined how language models can be supported in legal tasks by accessing relevant judgments. This work gave rise to the question that still occupies him today: how must legal information be prepared and linked so that AI can derive reliable answers from it?

With LexGraph, he now links more than one million court decisions with statutes, legislative materials and legal concepts in a knowledge graph. The focus is on the legal relationships between the sources: which provision a decision interprets, which line of argument it supports and in which temporal context it should be placed. Via API and MCP, this context is made available directly within legal AI products and existing AI workflows.

As CEO, Permantier is responsible for strategic direction and product development. His areas of expertise are agentic systems, legal research and the connection of language models with structured knowledge. He focuses on how AI finds relevant sources, takes legal connections into account and prepares results so that users can verify them against the original sources.

His work follows the principle of many small, measurable improvements. Together with his team, he continuously develops the data structure and search further, incorporating the experiences of law firms and legal tech companies. For him, what matters is whether the information provided helps to correctly answer a specific legal question.

Beyond his technical work, he brings experience in organising public events and communicating with diverse stakeholders. In his professional exchanges, he focuses in particular on the role of data quality, traceable research and open interfaces for the future of legal work.