We're building an Artificial General Intelligence for Law.
We are developing a jurisprudentially-grounded neuro-symbolic legal AI, with the aim of handling most legal problems in most jurisdictions. We have been working on it for over a year and hope to make it available in 2027.
Purpose
Senior lawyers have long been warning that the complexity of modern law has grown beyond the structural capacity of the legal profession, threatening the rule of law itself.
Our research aims to provide accurate legal reasoning, with appropriate awareness of arguments, gaps, indeterminacies, and interpretations. We believe this approach could help to address modern law's structural problem.
Approach
We combine frontier machine learning with symbolic architectures grounded in jurisprudence, logic and computational law. Our architecture is informed by over a hundred years of research into the normative structures underlying legal systems.
We embrace the complexities, nuances and messiness of the law in action, as well as the responsibility of professional integrity and reliability on the providers of legal services.
Ethics
AI threatens one form of the lawyer, not lawyers themselves. The legal profession has remade itself many times in the last two thousand years and can do so again.
Legal ethics and AI safety can be seen properties of a sociotechnical system that makes lawyers more responsible and responsive to the rule of law, not less.