A world-first German AI-based recommendation system for lawyers

Lawyers, judges, and other legal professionals must build a well-rounded picture of a case by looking at related incidents, making sure that no precedent escapes attention. Depending on the complexity of the case, this can require searching hundreds or even thousands of documents – a considerable investment in time and resources.

Manz, a legal publishing house based in Austria, leveraged deepset Cloud and Amazon Bedrock to create a solution that would transform their legal research by integrating AI to assist legal professionals in new, impactful ways, while also maintaining data accuracy and relevance.

The Challenge: information management

Manz’s online legal database, RDB Rechtsdatenbank is a paid offering that assists legal professionals in handling cases. Home to over three million documents, from legal rulings and contractual clauses, to other published content, it is updated on a daily basis. 

The basic search function in RDB lets users input a query and then returns all documents from the database that match the search terms. However, as Alexander Feldinger, product manager at Manz, explained: "usually one document doesn’t answer your question. It’s important to look at all the relevant documents — you don’t want to miss anything because that could have a negative effect on your work."

The Solution: MANZ Genjus KI

Manz's software development and product management teams partnerd with deepset and AWS to deliver a groundbreaking solution using deepset Cloud and AWS Bedrock, specifically leveraging the Anthropic Claude large language models. The result is a legal copilot—"MANZ Genjus KI"—an AI-powered assistant designed to safely enhance and accelerate legal research and document management. The solution focuses on two main use cases:

1. Text Similarity: This involves embedding and vector databases to improve the accuracy of legal research by identifying and retrieving similar documents more efficiently.
2. RAG Chatbot: A retrieval-augmented generation chatbot has been developed, capable of answering complex legal inquiries and iterating discussions based on historical queries, using pre-processed proprietary and open source data.

In order to ensure data quality and consistency, which is particularly important due to the complexity of legal documents and publications, the MANZ developers had to create a “knowledge base” from the existing and vast content data pool. The contextual, legal understanding of the content had to be combined with AI knowledge in order to guarantee the highest quality of the answers received.

Customer Benefits

With MANZ Genjus KI, MANZ is significantly improving the efficiency and accuracy of legal research. The legal copilot allows legal professionals to conduct AI-powered searches, ask complex legal questions, generate summaries of legal decisions, and in the future, even create new contracts or clauses with minimal input. With 200 selected experts having already tested the product since Q2/24, MANZ plans to open it for testing by the broader legal community in autumn, with full product readiness by early 2025. The solution has already achieved a high accuracy rate with continuous improvements being made to the underlying data and meta-data, following a customer-focused delivery model of constant iteration and improvement.

The solution solves critical industry-specific requirements including:

• Transparency and References: For professionals in the legal industry, having transparency and easy access to underlying data is crucial. The solution allows users to read and click through to the full documents, providing the necessary references and ensuring that all legal research is well-grounded in the source material.
• State Management for Iterative Discussions: The legal copilot is designed to maintain state across iterative discussions, recording and referencing the history of questions. This feature is vital for legal professionals who need to track the evolution of their queries and ensure consistency in their research.
• Specialized Legal Nomenclature: The solution is fine-tuned with an embedding model specialized for legal terminology, ensuring that the AI understands and accurately processes the specific language used in legal documents. This specialization is essential for delivering precise and relevant results in legal research.
• Up-to-Date Legal Information: The legal copilot ensures that legal professionals always have access to the most up-to-date and relevant information on cases and precedents. Given that court decisions can change over time, it is crucial to avoid outdated content. The solution continuously updates its database to reflect the latest legal rulings, helping users avoid relying on obsolete information.