What is AlloyDB AI

AlloyDB AI provides a suite of integrated generative AI capabilities in AlloyDB for PostgreSQL and AlloyDB Omni databases. This includes high-performance vector search, an AI query engine, and natural language to SQL generation. Build enterprise-ready AI applications faster and with less complexity by bringing AI to your operational data.

Learn more about AlloyDB AI

  • Vector search thumbnail
    Vector search

    Store, index, and query vector embeddings at scale using standard SQL operators. Find exact and approximate nearest neighbors with low latency. Fully compatible with PostgreSQL's pgvector extension.

  • AI Query Engine thumbnail
    AlloyDB AI query engine

    Use AlloyDB AI operators (Preview) in standard SQL queries to filter, rank, or generate multimodal content. Connect AlloyDB AI to Vertex AI foundation models and custom model endpoints using standard SQL functions. Generate embeddings or invoke predictions directly from your query without data movement.

  • Natural Language in SQL Queries thumbnail
    AlloyDB AI natural language

    Use everyday natural language prompts to generate executable SQL queries against your database schema. Empower analysts and developers to query data faster using AlloyDB AI natural language (Preview).

Documentation & resources

Explore guides, conceptual explanations, and references to dive deep into using AlloyDB AI to build AI/ML applications. These resources cover the complete workflow guides such as creating high-performance ScaNN indexes and performing searches to ranking results, backed by technical references such as model endpoint management and monitoring vector index metrics.

Customer stories

Learn how customers use AlloyDB AI to build generative AI applications.

Latest Videos

Watch the latest videos about AlloyDB AI.

Blogs & codelabs

Explore how AlloyDB AI is used in real-world scenarios like Target's search overhaul, dive deep into vector search enhancements and new innovations. Learn about hybrid search, multimodal capabilities, AI operators, and embeddings through videos and codelabs.