EnterpriseDB Corp., which sells a commercial version of the popular open-source PostgreSQL database management system, is launching itself into the artificial intelligence fray with today’s introduction of EDB Postgres AI.

It’s a platform that combines transactional, analytical and AI workloads — a single PostgreSQL package that can be deployed in a cloud instance, on-premises or via EDB’s managed cloud service formerly called BigAnimal.

“Fundamentally, EDB going through a transformation from being a Postgres database company to a Postgres data AI platform,” said Jozef de Vries, the company’s chief product engineering officer. “We are extending the power of Postgres to adjacent workloads where there is a tremendous amount of untapped value.”

Like many established vendors, EDB wants to assert its relevance in a market that has given birth to hundreds of startups.

Enterprises today “are primarily looking at new vendors and new technologies,” deVries said. “That introduces additional complications into their data state ecosystem. It means spinning up new database systems, moving data around, upskilling and building talent. If we can bring those solutions into the Postgres ecosystem, it helps simplify their outcomes.”

The new release adds a capability for users to conduct analytics on transactional data through a lakehouse that runs adjacent to the production database. The lakehouse can be configured as a cluster that synchronizes data and enables users to operate on it with the familiar Postgres interface.

The technology combines open-source technologies, including Apache DataFusion, Apache Arrow and Databricks Inc.’s Delta Lake storage framework. Support for Apache Iceberg will be added soon, deVries said.

New observability features allow administrators to manage both on-premises and cloud databases from a single pane of glass. It addresses the problem that “it has become more difficult to ensure that all database systems are running at the most current version level and that the health across that estate working as intended,” deVries said.

The unified observability capability requires installing an agent on each monitored database instance. It’s currently available for self-hosted databases, EDB’s managed service and cloud-managed and Amazon Web Services Inc. RDS PostgreSQL instances in the Amazon cloud. Support for other cloud platforms is planned.

This release also supports vector capabilities using the open-source pgvector extension. Vector databases are widely used in large language model training. EDB is making a preview version available that extends pgvector capabilities to support workflows for ingesting, searching and retrieving vector data.

“This ultimately is setting the stage for users being able to build their chatbot or information retrieval application entirely on top of a Postgres system provided by EDB,” deVries said.

EDB is also adding its own generative AI features by introducing an Oracle Migration Copilot that can assess the compatibility of an Oracle Corp. schema with Postgres.

Oracle Compatibility Mode has been a feature of PostgreSQL for years, but “when users run the compatibility assessment, they generally get to within the mid-90% range of compatibility before theu encounter things they have to rectify,” he said. The Migration Co-pilot uses OpenAI’s LLM supplemented by years of knowledge from EDB’s libraries.

“A user who has an error or an incompatibility issue can interact with the co-pilot in context to understand how to address the issue,” deVries said.

Image: EDB