We're pleased to announce the launch of Snowflake Notebooks in public preview, a highly anticipated addition to the Snowflake platform tailored specifically to integrate the best of Snowflake within a familiar notebook interface. Snowflake Notebooks aim to provide a convenient, easy-to-use interactive environment that seamlessly blends Python, SQL and Markdown, as well as integrations with key Snowflake offerings, like Snowpark ML, Streamlit, Cortex and Iceberg tables. You can now use Snowflake Notebooks to simplify the process of connecting to your data and to amplify your data engineering, analytics and machine learning workflows.

Keep all your data and development workflows within Snowflake’s security boundary, minimizing the need for data movement. Leverage your existing role-based access controls (RBAC) to manage access to notebooks and the underlying data assets to enable consistent and robust data governance. 

Notebook usage follows the same consumption-based model as Snowflake’s compute engine. This allows you to be charged only for what you use and gives you flexibility to use elastic compute options within Snowflake's single engine, for optimal price-for-performance.

Explore, analyze and visualize data using Python and SQL. Discover valuable business insights through exploratory data analysis. Train and manage your AI/ML models directly in your notebook. Develop scalable data pipelines and transformations for data engineering. Plus, leverage AI-powered editing features for faster, more efficient development.

Use native Git integration to version-control and collaborate on the same notebook files. Connect your preferred platform (GitHub, GitLab, Bitbucket, Azure DevOps) to manage and track changes for collaborative development. Schedule data ingestion, processing, model training and insight generation to enhance efficiency and consistency in your data processes.

To see what’s possible and change how you interact with Snowflake data, check out the various use cases you can achieve in a single interface: 

• Integrated data analysis: Manage your entire data workflow within a single, intuitive environment. Access Snowflake platform capabilities and data sets directly within your notebooks. This unified approach eliminates context switching and offers a way to transform raw data into actionable insights faster than ever before. See the possibilities below:
  • Access your Snowflake data directly — tables, stages and more — simplifying data pipeline development. No data movement or connection setup is required, thanks to native integration.
  • Integrate with Snowpark to make connection, authentication and interactive Snowpark dataframe display seamless and easy to set up. 
  • Create and run interactive Streamlit data apps directly within a notebook. 

• Build data pipelines and manage data: Simplify data engineering workflows, including data profiling, ingestion, transformation and orchestration. Automate deployments using Python APIs, Git integration and Snowflake CLI. Manage your databases like a pro — organize, store, manipulate and keep your data up to date for ready use.

• AI/ML development made easy: Build end-to-end machine learning pipelines with no additional development infrastructure and with the ease of use that you are accustomed to with Snowflake and Snowpark ML, all within Snowflake Notebooks. Preprocess data, perform feature engineering, create and train models using a curated set of ML packages, visualize results, and deploy and manage models — with end-to-end governance and no data movement or silos.

"Snowflake Notebooks help accelerate ML workflows. The seamless integration of experiment tracking with Weights & Biases directly within Notebooks eliminates context switching and streamlines the entire machine learning lifecycle for building and deploying models. We're excited to see how this integration unlocks further efficiency gains for our customers." 

—Venky Yerneni, Manager, Solution Architecture, Weights & Biases

For advanced ML workflows, Snowflake Notebooks now offers a new runtime option: Container Runtime (private preview). More flexible and powerful, Container Runtime provides seamless access to distributed processing with CPU and GPU options, ideal for resource-intensive machine learning tasks. Take advantage of a preinstalled, Snowflake managed set of popular machine learning packages to ensure compatibility and ease-of-use with zero manual setup, or pip install any custom package of choice. Container Runtime also includes optimizations for loading data stored in Snowflake, out-of-the-box distributed training support, automatic lineage capture and Snowflake Model Registry integration. The best part? Container Runtime is accessible at the click of a button, no manual setup or maintenance required.

Integrate your Git repositories with read/write access from supported platforms (GitHub, GitLab, BitBucket, Azure DevOps) as the source of truth across projects. This improves collaboration and helps enable teams to always work on the latest versions and are able to easily track changes. 

Go from experimentation to deployment all in one place. Automate and schedule your notebooks, simplifying large-scale data pipeline development. Seamlessly monitor progress by tracking the history of scheduled tasks and revisiting past runs in read-only mode for result inspection.

Use the scheduler to:

• Automate data pulls and transformations on a schedule
• Set up models to train and generate inferences automatically, according to your needs
• Generate reports and data summaries at regular intervals
• Automate model transformations and storage outputs
• Automate CI/CD workflows (see our recommended approach in this article)

With external access integrations, you can visit accepted network endpoints to enrich your workflows with additional data and packages (via the `!pip install` command in Container Runtime). External accesses are controlled on a per-notebook basis to preserve security and allow maximum flexibility at the same time. 

Accelerate your coding and data workflows with Snowflake Copilot — our breakthrough AI-powered SQL assistant now available directly within Snowflake Notebooks in select regions.

Snowflake Copilot can assist with:

• Writing SQL queries based on your data
• Improving your SQL queries through conversation
• Exploring your data with open-ended questions
• Learning about Snowflake concepts and features

Snowflake Copilot inline actions are also coming soon. Improve and learn more about your code. Select one of your SQL statements directly in a Notebook cell, and Copilot will allow you to optimize, explain or ask follow-up questions about it. Copilot will respond to you with a contextualized answer.

While we're introducing Notebooks as a powerful new way for data teams to accelerate their workflows, we also recognize the need for flexibility and choice. That's why we partner with Hex, to provide data teams with access to best-in-class tools to drive more decisions for the business and more value from their data. 

We invite you to explore Snowflake Notebooks and discover how it can enhance your data workflows. Stay tuned for additional enhancements, including support for complex ML workflows, additional options for runtime support, improved collaboration and sharing capabilities, and telemetry and debugging capabilities, just to name a few.

Your feedback and support are invaluable as we continue to refine and expand Snowflake Notebooks, so please try it today, directly in Snowsight, or sign up for a free trial account and help us make it better for you.

Get started with our documentation and explore the power of Notebooks firsthand. See it in action on our YouTube playlist, and explore sample notebooks on GitHub.

Snowflake Notebooks is now available in Warehouse Runtime (PuPr) for all Snowflake accounts deployed across AWS, Azure and GCP. Container Runtime (PrPr) access is currently limited to AWS, but we're rapidly expanding availability.