Key results
The company
Predori
predori.comAI-powered patent intelligence and monitoring platform for IP and R&D teams.
Result highlights
- 10x operational cost reduction
The story
A patent intelligence platform providing strategic insights on M&A and portfolio analysis for diverse industries, including the automotive sector.
The team initially built semantic search capabilities on a Lucene-based AWS service that required keeping all data in RAM to function. This setup was inefficient for vector retrieval, forcing the company to over-provision hardware and incur unsustainable infrastructure costs.
Predori migrated to Weaviate Cloud as a fully managed vector database to eliminate the operational burden of hardware management. The transition functioned as a drop-in replacement by simply swapping the legacy connection for the Weaviate Python client. This scalable foundation now supports the development of advanced agentic RAG systems beyond basic search capabilities.
Quotes
“We were trying to squeeze a vector database into what is not really a vector database, and it was super expensive and just not good. It was just the wrong technology, a complete overkill for the mathematical problem of approximate nearest neighbor search.”