Converting SQL¶

Converting SQL is turning source-platform DDL, views, procedures, functions, and scripts into Snowflake SQL (and the matching Snowflake objects) so you can deploy and test on Snowflake. The Snowflake Migration Agent drives this conversion: it runs SnowConvert AI for deterministic translation first, then layers AI-assisted remediation, assessment, and deployment on top of the same engine you can also reach from the SnowConvert AI CLI and desktop app linked at the end of this page.

SQL conversion in the Snowflake Migration Agent¶

The skill treats SQL conversion as a defined stage after your source is registered in the project, with deterministic translation as the baseline and AI as the remediation layer:

Capability	What it means
Register code units	Pull DDL and procedural code from a live SQL Server or Redshift connection, or import existing `.sql` files into the project
Convert code	Run SnowConvert AI to translate source SQL (for example T-SQL or Redshift SQL) into Snowflake-compatible SQL, with a per-object summary plus an EWI report (errors, warnings, informational issues)
AI-assisted conversion	After the deterministic pass, AI explains remaining EWIs, suggests fixes, and applies them interactively against the same converted code
Rule engine	Capture recurring fixes as reusable rules (regex or AI-guided) and propagate them across many files when the same EWI repeats

The conversion stage feeds directly into Stage 6 (Migrate), where deploy, data copy, testing, and iterative fix loops continue.

Example prompts (from the skill topic):

"Extract objects from the source"
"Import SQL files from ./my-scripts/"
"Convert my source code"

Use "What is the current state?" at any time to reopen the progress checklist.

Platform availability for code extraction, deterministic conversion, AI conversion and verification, and deploy appears in the Supported source systems table in the Snowflake Migration Agent (deterministic conversion is available across SQL Server, Redshift, Teradata, Oracle, and other dialects today; extraction, AI verification, and deploy currently target SQL Server and Redshift, with additional sources on the roadmap).

If you want to explore further, the SnowConvert AI documentation for SQL conversion is where you will find fuller answers on architecture and how to use the products: which dialects are supported and to what depth, how DDL and procedural code map to Snowflake (including Snowflake Scripting), how EWIs are organized by code, and how to remediate issues that the deterministic engine could not fully resolve.

Supported languages is the place to start for dialect-by-dialect coverage. Issue resolution is the matching deep dive for EWI codes, conversion issues, and functional differences. For product-wide terminology (EWIs, FDMs, AI code conversion at a high level), see SnowConvert AI — About.

If you are not driving the migration through Cortex Code, you can use the same SnowConvert AI engine from the SnowConvert AI CLI (scai) or the SnowConvert AI desktop user guide.