Case Study: Garden achieves sub-100ms patent intelligence and infringement analysis with Qdrant

Garden scales patent intelligence to 200M+ patents with Qdrant

Garden, a New York-based startup applying AI to patent intelligence, faced significant challenges scaling its platform. Their system, which required comparing millions of patent-product pairings in seconds, was hampered by their previous vector search solutions. These solutions lacked native filterable HNSW, forcing them to maintain separate indexes for every query combination, which was costly and inefficient. This outgrew their first fully-managed service and introduced operational pains with a subsequent self-hosted alternative.

By implementing Qdrant Cloud with its filterable HNSW and managed Rust backbone, Garden found a solution. The migration was straightforward, and Qdrant's pay-for-what-you-use pricing and scalar quantization allowed them to manage 10x more data for the same cost. The results were transformative: query latency dropped to under 100ms, the addressable patent corpus scaled from 20M to over 200M, and the new filtering capability unlocked an entirely new revenue line for high-confidence infringement detection.

Garden