Cortex Code CLI Expands Towards Supporting Any Data, Anywhere
Earlier this month at BUILD London, we announced the general availability of Cortex Code CLI, designed to transform the way data practitioners use and build on the Snowflake AI Data Cloud.
Today, we are excited to announce that Cortex Code CLI, in addition to supporting Snowflake workflows end to end, is now expanding support towards all data systems, starting with data engineering systems dbt and Apache Airflow®1. It allows developers to build transformations faster, optimize orchestration workflows and debug with full context.
We are also introducing Snowflake’s first self-serve subscription to help developers quickly get started with Cortex Code CLI and work with any data, anywhere.
Agentic data engineering for modern data stacks
Modern data stacks span multiple systems, which makes data engineering workflows particularly complex and resource intensive. When pipelines break, they create massive hidden costs through lost productivity and eroded data confidence.
The challenges for data engineers are clear: Creating, managing and optimizing data pipelines at scale requires juggling multiple complex tools. While Snowflake offers powerful native solutions such as Dynamic Tables and Snowpark, many organizations use a diverse ecosystem of tools, including dbt for transformation and Apache Airflow for orchestration. Managing this multisystem ecosystem manually is hard and demands significant expertise.
To address these challenges head-on, Cortex Code CLI now understands and supports popular dbt and Apache Airflow patterns, providing teams with optimized skills with a deep understanding of schemas, dependencies and orchestration logic using natural language.
With Cortex Code CLI, teams can:
Easily discover the data they need: Ask "Which tables contain revenue?" and get role-based-access-control-filtered results, dbt lineage and live statistics — not stale code.
Build and optimize faster: Generating dbt models is as simple as telling Cortex Code CLI to "create a CLV model joining orders and customers." Users can also auto-diff outputs, debug with full context and get fixes that actually compile.
- Perform production-ready orchestration: Build Airflow DAGs with dependency awareness, and debug from task to query.
"Cortex Code is transforming how we approach agentic analytics at Braze. Its native understanding of our data sets, schemas and columns means our engineers spend less time wrestling with context and more time getting precise, actionable outputs. We're deploying it against more complex data integrations and using it to automate and enrich our insights layer, turning structured analytics into something more dynamic, expressive and powerful."
Spencer Burke
Our rigorous testing demonstrated that Cortex Code is more accurate and efficient than generic coding tools in tasks requiring deep awareness of your data stack:
Higher accuracy: We ran Cortex Code and Claude Code against the set of ADE-Bench tasks using dbt + Snowflake with the same underlying model – Claude Opus 4.6. Cortex Code successfully completed 28 of 43 tasks (65%), compared to 25 of 43 (58%) for Claude Code.
Greater efficiency: Cortex Code required nearly 50% fewer total calls to reach the correct solution, including 2x fewer file reads and 4x fewer bash commands. This efficiency is driven by a SQL-native approach using snowflake_sql_execute tool, which enables native execution within Snowflake, resulting in more target exploration.
This is just the beginning — we're committed to expanding support for additional data engineering tools in the future.
"Cortex Code has transformed solution development at evolv Consulting by providing a direct, context-aware connection to the Snowflake ecosystem, allowing our team to interact seamlessly with databases, objects and git repositories. By leveraging the most advanced models available, such as Claude Opus 4.6, Cortex Code can essentially 'do anything' through its CLI version, from full-featured React app development to complex data engineering tasks. This translated into over 500 hours in time saving — roughly $100,000 in value — in just the first 20 days of adoption.”
Trent Foley
Expanded model choice to match your workload requirements
We understand that every organization has unique workload requirements and priorities when it comes to model selection. To provide Cortex Code CLI users with comprehensive choice without needing to look elsewhere, we're expanding our model offerings to include OpenAI GPT 5.2. This addition complements our existing suite of frontier Anthropic models — Claude Opus 4.6 and Claude Sonnet 4.6 — giving users the flexibility to optimize for quality, latency, cost and governance based on their specific needs.
Enterprise controls and governance for trusted AI
As we expand access, we remain committed to offer security and governance capabilities that enterprises require. Our new administrative control and governance features provide robust admin configurations to manage access, usage policies and permissions in the Cortex Code configuration. This enables developers to accelerate their work with AI-powered assistance while supporting organizations’ compliance with their security standards, regulatory requirements and operational governance frameworks.
CLI subscription: A new way to get started
Cortex Code CLI represents a fundamental shift in how data teams work — bringing intelligent, context-aware assistance to every part of your data stack, whether you're building on Snowflake, dbt or Apache Airflow or across multiple systems. With expanded model choice and enterprise-grade governance features, there's never been a better time to experience agentic data engineering firsthand.
To make Cortex Code CLI accessible to every developer, we're introducing a new low-friction subscription model. It means that non-Snowflake customers can start using Cortex Code CLI immediately.
If you already have a Snowflake account, learn more and get started with Cortex Code CLI documentation.
1 "Apache Airflow” is a registered trademark or trademark of the Apache® Software Foundation in the United States and/or other countries.
Authors
