Vespa

Vespa is an open-source search platform that combines full-text search, vector search, filtering, and ranking.

Vespa is Productboard's strategic search platform. The search experience powered by Vespa is now generally available to customers, and the platform has been validated in production. Use Vespa as the default for new keyword, vector, and hybrid search use cases rather than introducing new usage of Weaviate or OpenSearch.

Vespa supports tenant-defined custom-field filtering, high-frequency updates, and queries across multiple entity types with server-side ranking. Its tensor support keeps multiple chunk vectors in one entity document without duplicating metadata, simplifying the application and improving memory efficiency at scale.

See ADR-800: New Engine for Search for the decision, evaluated alternatives, and constraints, including latency limits for very large QNG ID lists.