Google’s AlloyDB AI Transforms Databases To Power Generative AI Apps

Google has introduced AlloyDB AI, an integrated set of capabilities within AlloyDB for PostgreSQL, aimed at supporting developers in building generative AI applications using their own data. With its built-in support for vector embeddings, AlloyDB AI provides the foundation for AI search apps and more. This new feature is available in preview through AlloyDB Omni, which is transitioning from technical to public preview.

AlloyDB AI, designed with portability and flexibility in mind, allows developers to incorporate real-time data into generative AI applications. Furthermore, it not only offers PostgreSQL compatibility but also enables customers to leverage AlloyDB AI through AlloyDB Omni to build AI-enabled applications in various environments, such as on-premises, at the edge, across clouds, or even on developer laptops.

Vector embeddings, numerical representations of data like text, audio, and images, play a crucial role in helping AI algorithms understand the relationships between different types of data and their semantic relevance. For instance, recommendation engines can utilize embeddings to find data similar to other data, such as similar movies or TV shows. However, the use cases for vector embeddings extend beyond that to areas like fraud detection and typo correction.

AlloyDB AI helps users transform data within databases into vector embeddings with just a single line of code, eliminating the need for a specialized data stack. While PostgreSQL already supports vectors, AlloyDB AI takes it a step further by providing access to Google’s on-premises embeddings models for in-database embeddings generation, as well as cloud embeddings models served through Vertex AI, Google’s platform for building and deploying AI applications.

Both the on-premises and Vertex AI models can generate embeddings on the fly in response to user inputs or automatically create embeddings through inferencing in any generated database columns. In addition to its models, AlloyDB AI offers up to 10 times faster vector query performance than standard PostgreSQL, thanks to its tight integrations with the AlloyDB query processing engine.

Furthermore, AlloyDB AI is integrated with Vertex AI Extensions, a set of fully managed tools that assist developers in connecting models to proprietary data or third-party sources. It also supports LangChain, an open framework that simplifies the development of apps leveraging generative AI text models.

AlloyDB AI is currently available through AlloyDB Omni and will be launched later this year on the AlloyDB managed service. Users can add the capabilities of AlloyDB AI to any AlloyDB deployment by installing the relevant extensions, with no additional charge.