Cortex Code extends capabilities: a unified managed agent for Your Entire Data Stack

Snowflake has announced an update to its Cortex Code tool, which now offers a single managed AI agent for developers' full data stack, including platforms such as Snowflake, dbt, Airflow, Databricks, and AWS Glue. This update significantly simplifies data operations and accelerates the development process in complex information management environments.

Originally launched this spring as an AI coder specifically adapted for Snowflake, Cortex Code has now become a multifunctional tool that integrates with various systems and supports existing developer tools. Among the new capabilities are visual intelligence and enhanced execution features, accessible via the Snowsight interface.

These enhancements reduce the time spent on context restoration and error resolution when working with different tools, which is particularly important for modern teams using diverse technologies. Simplifying interoperability between systems can improve productivity and data quality.

Additionally, new Cortex Code features allow development teams to quickly adapt and scale in a rapidly changing data market. Optimization of workload translations and dbt project version management provide greater flexibility and control, which is critically important for increasing operational efficiency.

The strongest part of the announcement is governance. Snowflake is positioning Cortex Code as an agent that can work across the stack while still respecting enterprise controls, metadata, and access boundaries. That distinction matters because data teams usually cannot hand production pipelines to a generic coding assistant without auditability, permission checks, and context about how assets are connected.

For teams already juggling Snowflake SQL, dbt models, Airflow DAGs, Databricks notebooks, and AWS Glue jobs, the value is less about writing isolated snippets and more about reducing context switching. A managed agent that understands the surrounding workflow can help explain failures, suggest safe changes, and keep implementation decisions closer to the systems where the data actually lives.