pgEdge adds decision models to its open source Postgres monitoring tool

Nov 19, 2026
In pgEdge's test, TypeSafe AI's Jev agreed with Claude Haiku 4.5 on 89% of alerts at about one-fiftieth of the cost.

pgEdge has added support for decision models to its open source AI DBA Workbench, a tool that monitors Postgres databases and flags problems while keeping a human administrator in the loop on fixes. The support, in the new 2.0 beta, lets the Workbench use TypeSafe AI's Jev and OpenAI models accessed through the OpenAI Decisions API to sort database alerts.

In pgEdge's own test on 30 days of monitoring data, Jev agreed with Claude Haiku 4.5 on 89% of alerts and never suppressed an alert that Haiku had flagged. It cost $0.04 per 1,000 decisions against $1.93 for Haiku, and took 200 milliseconds against three seconds. pgEdge says this lets organisations put off by token costs adopt agentic Postgres monitoring without high AI bills. Because Jev gives no explanation, the Workbench can pass an alert to a full language model when an administrator wants one.

"Adding the ability to fall back to a full LLM for reasoning about specific alerts is a very powerful combination and applies the right model to the right task," said Dave Page, pgEdge's chief technology officer. The Workbench works with standard Postgres from version 15, the company said in its release. Jev also powers release decisions in Pulumi's new delivery tool.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like AI Programming Weekly or Daily AI Brief.

Also, consider following us on social media:

Free newsletter

AI Programming Weekly

Weekly news about AI tools for software engineers, AI enabled IDE's and much more.

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more
Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.