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

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

For the fourth straight year, Stack Overflow’s Developer Survey has found that PostgreSQL tops the list of the database management systems developers prefer — and the favorability ratings even ticked up from last year.

If it seems odd that a 40-year-old open-source engine should have such staying power, Phillip Merrick (pictured), co-founder and chief executive of pgEdge Inc., says the reason is simple: The rapid rise of artificial intelligence coding agents is giving PostgreSQL new life. “It’s the default database for agent-generated applications,” he said in an interview. “Demand has skyrocketed.”

That’s important, since AI agents now spin up 80% of enterprise databases, according to Databricks Inc. PostgreSQL provides a common foundation for building applications, retrieving information for language models and helping administrators manage the resulting database sprawl.

AI has introduced new kinds of database features to the development mix, including vectors, graphs and agentic branching. PostgreSQL’s reputation for stability and performance, along with its extensible architecture, appeals to developers overwhelmed by new database architectures.

“A lot of people don’t realize that this explosion in agentic application development is creating an explosion in the number of databases to manage,” Merrick said. “You’ve only got a fixed number of human [database administrators].”

A key appeal of PostgreSQL is its extensible architecture, which lets developers add AI capabilities without adopting an entirely separate database engine. For example, the pgvector extension allows vectors to be stored alongside other data and used to search for similar items, key capabilities in retrieval-augmented generation. PostgreSQL’s open-source code and publicly available documentation also make it accessible to coding agents, which can use it to build more applications.

Keeping application records and vector representations within PostgreSQL reduces the number of database technologies teams must learn and operate. It also gives retrieval workloads access to familiar database features such as transactions and joins.

Vendors such as pgEdge are building their businesses on developers’ need to balance rapid experimentation with the security, operational control and deployment flexibility enterprises demand. The company’s recently announced Starfleet platform is intended to address demand for infrastructure that makes an application easy to build and support reliably in production.

Merrick asserts that standardization can lower training and management overhead. But simplifying the technology portfolio doesn’t necessarily simplify the workload. That creates a second role for AI: helping database administrators diagnose problems and propose repairs.

PgEdge’s AI DBA Workbench monitors databases, identifies issues and generates SQL to address them with the approval of human administrators. Merrick cited the example of troubleshooting a report whose runtime has suddenly stretched from minutes to hours. An agent might identify a configuration change or missing index, but investigating the problem shouldn’t give it unrestricted authority to alter the database.

PgEdge’s SafeSession technology enforces read-only access at the PostgreSQL engine level. Its Model Context Protocol server connects agents to database contents and schema information, making access controls part of the integration itself. An agentic database administration platform called AI DBA Workbench “monitors and manages your database, finds problems and diagnoses them correctly,” Merrick said.

Customers can build prototypes with StarFleet in pgEdge’s cloud and deploy using the same PostgreSQL distribution in managed cloud, customer-controlled cloud or on-premises environments, according to the company.

That flexibility addresses cases where data location or air-gapped infrastructure constrains production choices. It also accommodates the growing popularity of open-source and open-weight databases that businesses want to run on-premises or in a private cloud.

Merrick said PostgreSQL is becoming a meeting point for coding, retrieval and operations thanks to broad developer support.

“There’s a mantra among developers,” he said. “For just about any data management problem, the answer is Postgres.”

Photo: pgEdge