Customer Zero: How Retool Scaled 1 Million Databases on Databricks Lakebase
Overview
| Experience | In Person |
|---|---|
| Track | Lakebase |
| Industry | Healthcare & Life Sciences, Manufacturing, Financial Services |
| Technologies | Unity Catalog, Lakebase |
| Skill Level | Intermediate |
| DOWNLOAD SESSION SLIDES | |
Learn how Retool, an enterprise AppGen platform for building intelligent applications and agents, became an early adopter of Databricks Lakebase, creating over 1 million serverless Postgres databases to power its operational foundation. We’ll share the architecture, governance, and lessons learned as Retool scaled its own platform to support production-ready apps and agents that turn governed business data into action.
You’ll see how Retool’s Lakebase-powered operational layer now gives Databricks customers a proven path to run more workloads and remove data silos, whether they use Retool’s layer or connect directly to their own databases. The session includes a live demo to help you move from prototype to production with unified governance, automation, and lower operating costs through autoscaling.
Session Speakers
Abhishek Gupta
/Chief Product Officer
Retool
Full Summary
How Retool rebuilt its database layer on Lakebase to run a million serverless Postgres databases
AI apps and agents are spreading across organizations, often touching sensitive data and demanding transactional state. Retool reached a point where every new app or agent needed its own production-grade database, instantly provisioned, auto-scaling, governed, and cost efficient. Always-on Postgres clusters could not meet that bar at Retool's scale.
FAQ
Lakebase is the serverless Postgres service inside the Databricks Platform, powered by Neon. It delivers transactional Postgres that scales to zero, provisions via APIs, and inherits Unity Catalog governance automatically.
Managing over a million always-on Postgres databases created untenable cost, operational overhead, and governance risk. Lakebase offered serverless scale to zero, API-first provisioning in the app's code path, and native Unity Catalog enforcement, which changed the unit economics and removed the need for a dedicated infra team.
Compute stops when databases are idle, so Retool pays only for compute and storage actually used. Eliminating idle capacity translated to millions in annual savings that grow as usage increases.
Because Lakebase runs within the Databricks Platform, every database automatically inherits Unity Catalog lineage, policies, and role-based access controls. Enforcement happens at the data layer, so governance scales with app proliferation without extra configuration.
Reckitt reduced marketing teams' time to data by over 60 percent with the Lakehouse connector. Wave builds operational apps on fleet data using the Lakebase connector, deploying in minutes. MIQ is enabling agent-to-agent handoffs between Genie and Retool agents to trigger actions without manual data movement.