Databricks makes Lakebase Postgres generally available on Google Cloud

Databricks says Lakebase Postgres is now generally available on Google Cloud, where teams can run it on the Databricks Data + AI Platform. Lakebase separates compute from storage and shares its storage layer with analytical data, so teams do not need to build and maintain ETL pipelines to keep the two in sync.

The service offers serverless autoscaling with scale to zero, instant zero-copy branches of production data, and point-in-time recovery. It supports standard Postgres tools, drivers and extensions, and access is governed through Unity Catalog. KaBuM! is cited as using Lakebase Postgres as the default memory layer for its AI agents.

Multi-zone endpoints with automatic failover carry a 99.99% SLA, and single-zone endpoints carry a 99.9% SLA. Private connectivity through Google Cloud Private Service Connect is in beta for enterprise customers. The service is available in asia-southeast1, europe-west3, us-central1, us-east4 and us-west1.

Agent builders on Google Cloud can keep operational agent data in Postgres without maintaining sync jobs to their analytics store. Scale to zero means idle agent backends stop accruing compute charges.

Private Service Connect is still in beta, so teams that require private networking should check its status before depending on it in production.