CuratedAI is a Belgium based company that offers an AI assistant for privacy and IT law. They make EU legislation and case law searchable in natural language, down to the most relevant article or paragraph.
Industry
Legal Technology
Use Cases
Multi-lingual RAG
Cloud
AWS
Solution
Vespa.ai
Client Overview
CuratedAI is a legal technology startup specializing in privacy and IT law. The platform enables legal professionals to search through EU legislation and case law using natural language queries, providing relevant legal references and structured answers. Their platform is designed to enhance efficiency and productivity for legal professionals by simplifying the process of finding and utilizing legal documents and regulatory information.
Challenge
CuratedAI needed a multilingual search solution capable of retrieving legal documents across multiple languages. Users needed to query in English, French, or Dutch and receive results in the same language, even when official translations were unavailable. Additionally, the system had to support both broad exploratory searches and deep searches into specific legal documents, while ensuring that new legislation and case law were continuously integrated. Initially, CuratedAI's search system relied on a different vector database, but it faced limitations in handling multilingual queries and efficiently retrieving legal documents. A migration to a more scalable and flexible search engine was required to support hybrid search, combining both lexical and semantic search for better precision.
Solution
Searchplex migrated CuratedAI's search system to Vespa.ai, leveraging its multi-vector indexing capabilities to enhance multilingual search performance. The key components of the solution included:
- Migration to Vespa.ai → Transitioned from the previous vector database to Vespa.ai, allowing for more efficient document indexing and retrieval.
- Multilingual Embedding Models → Integrated multilingual embedding models to process and retrieve legal documents across languages.
- Hybrid Search → Combined lexical search for keyword-based retrieval with vector search for semantic understanding, ensuring more accurate search results.
- Multi-Vector Indexing → Used Vespa.ai's multi-vector indexing to capture different aspects of legal texts, improving retrieval quality for complex legal queries. This solution enabled precise and structured search results, aligning with the needs of legal professionals who require highly relevant legal references.
Result
The migration to Vespa.ai resulted in significant improvements to CuratedAI's search capabilities:
- Better multilingual search → Queries in English, French, and Dutch now return highly relevant results, even for documents without official translations.
- Improved legal document retrieval → Hybrid search and multi-vector indexing enhanced both broad and deep search experiences.
- Seamless transition → Migrating from the previous vector database to Vespa.ai provided greater flexibility for future expansions. With Vespa.ai powering its legal search, CuratedAI now delivers a more efficient and scalable retrieval system that meets the needs of legal professionals handling complex, multilingual legal research.