The Replit-Databricks integration is now generally available, with new native Lakebase support! Now, you can use Replit ⠕ to build full-stack Databricks Apps on live, governed enterprise data. Replit Agent can automatically provision a Lakebase database, while AI-proposed schema changes require human approval. Build, deploy, and iterate in one... flow. See it in action:show more

@Replit Governed enterprise data + human approval is a strong foundation for reliable AI apps. The next layer is making the context behind agent decisions equally traceable and evidence-backed.

@amasad @Replit generally? what do you mean generally?

@Replit The hard part of adding AI to an existing data stack is preserving the guardrails people already trust. Building into governed data changes the conversation from demo to operating system.

@Replit For schema approval, I’d want the proposal to show what existing queries and app paths might break, not just the SQL diff. “Add a required field” looks tiny until the old client tries to write a row.

Human approval for schema changes is the correct default. The same rule should apply to any write that changes the system of record: customer data, money, access, or production config. If the agent can mutate state without a named approver, the “governed data” layer stops at the prompt.

@Replit I think this is useful for enterprise data apps. The hard bit is still permissions, audit, and not letting AI agent changes bypass controls.