What it is: A set of production-grade retrieval features for Mastra's MongoDB integration: hybrid search that combines vector and full-text matching, Voyage rerankers to improve result relevance, automated embeddings that handle vectorization server-side, and pre-filters that work reliably at scale without hitting database limits.

Who it's for: Developers building RAG applications and AI agents with Mastra who use MongoDB as their vector database and retrieval backend.

Why it matters: Better retrieval quality directly improves the accuracy of AI answers that end users see. These features remove common friction points: manual embedding management, vector-only search limitations, and scale failures with large filtered datasets. This positions MongoDB as a complete, production-ready retrieval stack within Mastra.