Lexroom raises $50M in Series B funding to develop legal AI for European civil law

This round funds a groundbreaking thesis: legal AI must be built from the data layer rather than by refining general-purpose large language models (LLMs). Most legal AI products have been designed around a model in which a general-purpose LLM is fine-tuned and embedded into a legal interface.

The problem? Practicing lawyers cannot verify it, cannot cite it, and are increasingly refusing to use it. The risk is no longer theoretical: more than 1,300 court filings containing AI-generated inaccuracies — including fabricated citations, non-existent precedents, and misquoted authorities — have already been identified, recently forcing prestigious law firms to apologize to judges.

Lexroom has been designed with a data-first approach, built on proprietary infrastructure powered by more than six million verified legal sources — legislation, case law, and regulations — that are continuously updated and optimized for retrieval. Lexroom starts from the same place lawyers do: legal sources.

“When we founded Lexroom, two things immediately became clear: lawyers needed a better way to work, and LLMs could help make that possible. What was missing was the data — always up-to-date laws, relevant case law, and legal procedures. Civil law countries need an AI legal engine built around data-driven reasoning. We are excited to partner with the Left Lane team to build the backbone of AI in the legal industry,” said Paolo Fois, CEO and co-founder of Lexroom.

The architecture matters because of who uses it. Two-thirds of Lexroom users rely on the platform daily, while 94% use it weekly — engagement levels typically reserved for core workplace tools such as email and Microsoft Word. The product is transforming how lawyers work: research that once took hours is now completed in minutes, drafts that once required days are finalized in hours, and individual lawyers can serve more clients, faster, without compromising quality.

This multiplier effect extends across the entire legal system. Europe is home to more than one million registered lawyers, most of whom practice independently or within small firms — the segment that handles the majority of legal workloads and where most citizens first interact with the law.

When these firms move faster, the legal system moves faster as well: there is less delay between a client’s question and a substantiated answer, and less friction in the process that determines whether justice is delivered promptly. The 8,000 firms already using Lexroom operate on a fundamentally different cost and speed curve; the Series B funding round will finance the expansion of this transformation across Europe.

“Paolo, Martina, and Andrea have built an enterprise product with an exceptional user experience, something extremely rare in a sector that is often sold through innovation committees. We loved that Lexroom enables lawyers to build better businesses by providing more attentive and effective service to their clients, turning AI into an ally rather than an enemy,” said Paddy Dillon, Vice President at Left Lane Capital.

Italy, with approximately 250,000 registered lawyers and one of Europe’s most complex legal systems, served as the testing ground. The Series B round will fund expansion across Europe, beginning with Spain and Germany, supported by local teams and jurisdiction-specific capabilities developed in collaboration with firms already established in those markets.

About Lexroom

Lexroom is an AI platform built on proprietary infrastructure powered by millions of verified legal sources and a data-centric approach. Thanks to this architecture, Lexroom delivers reliable and fully traceable content back to its original source, enabling it to become the daily research, drafting, and analysis platform for more than 8,000 law firms and in-house legal teams. Founded in Milan in 2023 by Paolo Fois, Martina Domenicali, and Andrea Lonza, Lexroom is currently expanding across the European civil law market.

For more info, please visite lexroom.ai