Agentic SQL code converter is generally available in Databricks

The agentic code converter is now generally available (GA) for targeting Databricks SQL. Powered by Genie Code, it translates queries from legacy database dialects, automatically testing syntax and iteratively repairing errors before writing files.

Use cases

Imagine you need to migrate hundreds of analytical queries from Snowflake, T-SQL, Redshift, Oracle, BigQuery, or Teradata into Databricks. Doing this by hand requires hours of tedious syntax tweaks and dialect remapping.

With the agentic code converter, you can:

• Group scripts into workspace migration projects and process files concurrently.
• Generate ANSI-compliant Databricks SQL queries.
• Automatically validate syntax and fix dialect errors through autonomous subagents.
• Define custom skills to enforce internal type mappings and target standards.
• Run quick single-file translations directly using the /migrate chat command.

How to use

You can initiate migrations from migration projects in the workspace or run conversions directly in the Genie Code chat.

For a single file or directory, trigger the converter in the Genie Code chat panel:

/migrate

To enforce specific data type conventions across conversions, define a custom skill file at Workspace/.assistant/skills/sql-migration/SKILL.md:

---
name: sql-migration
description: Conversion rules to apply when migrating legacy SQL to ANSI SQL.
---

Always convert the custom type MYDATE to TIMESTAMP_NTZ.

What happens in this code:

1. The YAML frontmatter registers the skill name and scope within the workspace.
2. The plain-text instruction specifies the exact dialect conversion behavior to follow.
3. Genie Code subagents reference this rule across all converted scripts in the project.

Limitations and requirements

• Stage: GA for the SQL target framework. Generating Databricks notebooks (DataFrame API) is in Beta and requires workspace admin activation under Previews.
• Supported source dialects: T-SQL, Snowflake SQL, Redshift SQL, Oracle, BigQuery, and Teradata.
• Batch limits: conversions are limited to a maximum of 300 files per run.
• Unsupported: Teradata BTEQ scripts are not yet supported.
• Prerequisites: source .sql files must reside in workspace files, and the workspace must meet Genie Code prerequisites.