Snowflake's AI Coder Goes Broad

Snowflake is broadening the reach of its AI coding assistant, Cortex Code. The tool, designed to accelerate data engineering tasks, is now generally available to all Snowflake users directly within the Snowsight interface, according to Snowflake.

This expansion targets data practitioners of all skill levels, aiming to streamline the process from idea generation to production deployment and scaling.

Cortex Code in Snowsight

Cortex Code now serves as a persistent AI coding agent within Snowflake Workspaces and Notebooks. It offers context-aware assistance, understanding the user's current data catalog, including databases, schemas, and tables.

Users can generate, modify, and explain SQL and Python code conversationally. AI-suggested changes are previewable, and failed queries can be fixed with a single click.

The agent can accelerate end-to-end development, including building ML workflows and dbt pipelines, all through natural language prompts.

It also facilitates data discovery, allowing users to search for database objects using plain language. Answers about Snowflake features and best practices are drawn from official documentation and the Horizon Catalog.

Furthermore, the agent can handle administrative tasks, such as querying governance, security, and spend information conversationally. This includes identifying PII-tagged tables and monitoring credit consumption.

Expanded Developer Tools

Snowflake has also introduced native Windows support for the Cortex Code CLI. This allows developers to use the agent within their preferred editors like VS Code, directly on Windows machines.

The CLI simplifies and automates tedious development tasks across data systems, accelerating application and pipeline deployment.

Agent Teams and New Skills

A significant addition is the introduction of Agent Teams, transforming Cortex Code into a multi-agent orchestrator. This feature allows large tasks to be broken down and executed in parallel by independent sub-agents, each with specific roles.

A lead agent can coordinate specialized teammates for research, coding, and testing, managing task dependencies and output flow.

Users can intervene in live agent sessions to review progress or redirect work without losing context.

New agent skills further enhance Cortex Code's adaptability. These skills act as specialized playbooks, providing expert guidance for tasks like cost optimization, ML deployment, and Streamlit styling.

Skills ensure accurate, domain-aware responses without overwhelming the agent's general knowledge base.

Notable new skills include a guide for new users, tools for developing Streamlit apps, data movement via Openflow, cost intelligence for spend analysis, and a comprehensive machine-learning lifecycle skill.

The Cortex Code AI coding agent is now more accessible and capable of handling a wider array of data engineering and development tasks.

This initiative is part of Snowflake's broader effort to leverage AI for business users, as seen in Snowflake's AI coding agent developments.