Snowflake is broadening the reach of its AI coding assistant, Cortex Code. Initially designed with a Snowflake-centric view, the tool now aims to understand and operate across an entire data stack, including tools like dbt and Apache Airflow. This expansion addresses a key pain point for developers: AI assistants that lack the contextual awareness of data schemas, lineage, and access policies inherent in data-specific workflows.
The company announced that Cortex Code now supports additional data systems such as AWS Glue, Databricks, and Postgres. This allows data engineers to maintain context across disparate systems, reducing the time spent re-establishing context and debugging cross-system failures. The goal is a single AI agent that grasps the user's complete data footprint.
New Agent Skills for Data Engineering
Cortex Code is enhancing its capabilities with specialized agent 'skills' designed for core data engineering tasks. These include expert workflows for Snowpark Python, data quality checks, lineage tracking, and cost intelligence.
A new snowpark-python skill offers guidance throughout the Python pipeline lifecycle, from authoring code to deployment and observability. The snowpark-connect skill aims to streamline the migration of existing Apache Spark workloads to Snowflake, analyzing and fixing compatibility issues.
For those managing data transformation, the dbt-projects-on-snowflake skill provides native Snowflake object management for dbt projects. This includes deployment, versioning, documentation generation, and scheduling. Another skill, dcm, covers the full lifecycle of Declarative Configuration Management projects within Snowflake.
The daily brief
Get the free daily brief with the day's AI news, new AI startups and funding rounds every morning.
Join 64,000 other subscribers.
An Embeddable Agent Platform
Snowflake is transforming Cortex Code into an agent platform. With the Cortex Code Agent SDK, teams can program the agent using Python and TypeScript, embedding its capabilities into their own tools and workflows in a headless mode.
The AI coding agent is also becoming more accessible through integrations. A native VS Code extension is in private preview, and support for protocols like Model Context Protocol (MCP) and Agent Client Protocol (ACP) means it can integrate with other coding agents and editors, including Zed and JetBrains. This allows developers to leverage Cortex Code's data-native AI directly within their preferred development environments.
Enhanced Snowsight Experience
Within Snowflake's Snowsight interface, Cortex Code is now powered by Cloud Agents, offering expanded tool execution capabilities, including web browsing and file system access, within a governed cloud environment. This enables long-running tasks like pipeline builds and complex debugging sessions.
New user experience features in Snowsight include Plan Mode, which allows users to review and edit the AI's execution plan before it runs, fostering trust. Snap & Ask enables users to interact directly with visual elements like charts and DAGs within Snowsight, asking questions grounded in live data context without manual data description.
This push towards a more integrated and context-aware AI coding agent data stack aims to accelerate development cycles for data practitioners by providing intelligent assistance directly within their existing workflows. According to Snowflake, over 50% of their customers are already using Cortex Code.
