How agents are stoking new interest in a 40-year-old database

For the fourth consecutive year, a survey conducted by Stack Overflow has revealed that PostgreSQL continues to reign supreme among the database management systems favored by developers. The developer community's growing admiration for this 40-year-old open-source engine is attributed to the increasing popularity of artificial intelligence coding agents, according to Phillip Merrick, co-founder and CEO of pgEdge Inc. "It's the default database for agent-generated applications," Merrick explained.

"Demand has skyrocketed." With AI agents now powering 80% of enterprise databases, PostgreSQL provides a common ground for constructing applications, retrieving information for language models, and managing the resulting database sprawl. The extensive capabilities of PostgreSQL, including its stability, performance, and extensible architecture, make it attractive to developers overwhelmed by new database architectures.

The rise in agentic application development has resulted in a surge in the number of databases to manage, and PostgreSQL's extensible architecture allows developers to incorporate AI features without adopting an entirely separate database engine. For instance, the pgvector extension enables vectors to be stored alongside other data and used for similar item searches, which is crucial for retrieval-augmented generation.

Additionally, PostgreSQL's open-source code and publicly available documentation make it easily accessible to coding agents, allowing them to build more applications. By keeping application records and vector representations within PostgreSQL, development teams can minimize the number of database technologies they need to learn and operate.

They also benefit from PostgreSQL's familiar database features, such as transactions and joins, when dealing with retrieval workloads. Companies like pgEdge are capitalizing on developers' need to balance rapid experimentation with enterprise-level security, operational control, and deployment flexibility. The recently introduced Starfleet platform by pgEdge aims to address this demand by providing infrastructure that simplifies application building and reliable production deployment.

Standardization can reduce training and management overhead, but simplifying the technology portfolio does not necessarily simplify the workload. Therefore, AI plays a second role in assisting database administrators in diagnosing issues and proposing repairs. PgEdge's AI DBA Workbench monitors databases, identifies problems, and generates SQL to address them with the approval of human administrators.

To ensure safety, PgEdge's SafeSession technology enforces read-only access at the PostgreSQL engine level. The Model Context Protocol server connects agents to database contents and schema information, integrating access controls into the system itself. An agentic database administration platform called AI DBA Workbench can "monitor and manage your database, find problems and diagnose them correctly," according to Merrick.

Customers can build prototypes using StarFleet in pgEdge's cloud and deploy it using the same PostgreSQL distribution in managed cloud, customer-controlled cloud, or on-premises environments. This flexibility caters to cases where data location or air-gapped infrastructure restricts production choices. It also accommodates the increasing popularity of open-source and open-weight databases that businesses want to run on-premises or in private clouds.

Merrick emphasizes that PostgreSQL is becoming a meeting point for coding, retrieval, and operations due to broad developer support. "There's a mantra among developers," he stated. "For just about any data management problem, the answer is Postgres."