ShapedQL

ShapedQL is an end-to-end relevance engine that provides real-time personalization for search, recommendation feeds, and AI agent memory. It allows users to connect their data, train machine learning models, and query text, user, or session context to retrieve relevant results in milliseconds. The platform compiles SQL queries into optimized, multi-stage ranking pipelines, enabling hybrid search, hard constraints, ML model scoring, and reordering for diversity and exploration. ShapedQL is designed for product and engineering teams looking to enhance user engagement and drive revenue through personalized experiences. It offers a three-layer architecture with a query layer for real-time retrieval and ranking, an intelligence layer for ML models and embeddings, and a data layer with over 30 connectors for batch and streaming data. The platform boasts rapid experimentation capabilities, allowing teams to deploy and test new ranking models in days rather than months, and is built for enterprise scale with high reliability and security compliance. Key use cases include personalized content feeds ("For you" feeds), hybrid search and discovery, contextual memory for AI agents, similar item recommendations, personalized email content, and AI assistant recommendations. ShapedQL aims to replace traditional document retrieval systems with an engine that treats user context as a first-class input, offering faster deployment, instant updates, and the ability to learn from behavior automatically.