MyScale

MyScale: The SQL Vector Database for Scalable AI is a fully SQL-compatible vector database designed to empower developers to build production-grade GenAI applications with familiar SQL. It combines fast vector search, full-text search, and flexible metadata filtering with robust integration capabilities, enabling powerful Retrieval-Augmented Generation (RAG) use cases in a single system. The MSTG vector engine delivers high performance and cost efficiency, with strong security and compliance features to support enterprise deployments.

How to Use MyScale

1. Import data into MyScale from your data sources.
2. Create a vector index on your vector columns (e.g., using MSTG with a cosine metric).
3. Run SQL queries that combine vector distance metrics with traditional SQL operations (JOINs, filters, aggregations).
4. Leverage full-text search (BM25) and metadata filters to refine results.
5. Build RAG workflows by integrating vector search results with your LLM prompts and applications.
6. Monitor and observability via MyScale Telemetry to maintain production-grade visibility.

Core Capabilities

• SQL-compatible vector database for seamless adoption by SQL-savvy developers
• Fast, powerful vector search combined with text search and filtered search
• Complex SQL queries including vector joins and metadata filtering
• MSTG vector engine for improved performance and cost efficiency (3x speed and 3x savings claim in documentation)
• Flexible data modeling to manage various data types within a single system
• Deep integration with popular development languages and AI stacks
• Security and compliance features (SOC 2, ISO 27001) and SQL-based RBAC
• Rich observability through MyScale Telemetry for LLM application monitoring
• Built-in RAG support enabling improved knowledge retrieval and reasoning
• SDKs, integrations, and tooling to accelerate development (Langchain compatibility, etc.)
• Supported with managed cloud offerings (MyScale Cloud) and local deployment options

Feature Highlights

• SQL-native vector indexing with MSTG (metric_type=Cosine) for high-performance vector search
• Text search via BM25 and configurable tokenizers for multiple languages
• Vector + text search joint capabilities to optimize RAG results
• Advanced metadata filtering to precisely tailor result sets
• Easy data ingestion, indexing, and querying from a single system
• Comprehensive integration with popular languages and AI stacks
• Observability and monitoring through dedicated telemetry tooling
• Security: SOC 2 and ISO 27001 compliance, plus SQL-based RBAC
• Deployment flexibility with cloud and on-prem/offline options

Why Choose MyScale

• Familiar SQL interface lowers learning curve while unlocking vector search capabilities
• Unified platform for vector search, full-text search, and structured data
• Scalable, production-grade performance with cost-aware MSTG indexing
• Strong focus on integration, security, and observability for enterprise needs