A SQL-based tool for building and optimizing ranking pipelines.
ShapedQL is a declarative SQL-based interface and runtime that compiles SQL queries into optimized, multi-stage ranking pipelines for real-time relevance. It enables hybrid search across multiple indexes, hard-constraint filtering, ML-driven scoring, and diversity-driven reordering, powering search, recommendation, and personalized feeds. With 30+ native connectors and a three-layer architecture (query, intelligence, data), you can connect data, train models, and deploy ranking changes quickly with sub-50ms latency.
Hybrid search across multiple indexes
Filter with hard constraints & business rules
Score with ML models & value functions
Reorder with diversity & exploration
<50ms latency
Real-time retrieval and ranking
30+ native connectors