Weaviate 1.37 Release: MCP Server, Tokenizers, MMR, Query Profiling

AI Impact Summary

Weaviate 1.37 introduces several significant preview features designed to enhance LLM integration, text analysis, and large-scale operations. The built-in MCP Server allows LLMs to directly interact with Weaviate databases, shifting Weaviate from a passive retrieval engine to an active memory for agentic workflows. Furthermore, Extensible Tokenizers with accent folding and custom stopword presets improve text search quality, while Diversity Search (MMR) addresses the issue of redundant results in vector search. Finally, Query Profiling provides detailed timing breakdowns for queries, aiding in performance optimization.