A Breakthrough AI-Powered SQL Assistant

Posted on by

Seamless AI within Snowflake

Data is the lifeblood of modern businesses, but unlocking its true insights often requires complex SQL queries. These queries can be time-consuming to write and challenging to maintain. At Snowflake, we believe in making the power of data accessible to all. That’s why we prioritize simplicity, governance and quality in everything we build — including our AI-powered tools. We’re thrilled to announce that Snowflake Copilot, a new solution on the bleeding edge of text-to-SQL that simplifies data analysis while maintaining robust governance, is now generally available. 

Snowflake Copilot: Your AI-powered SQL assistant is now generally available

 Leveraging breakthroughs from Snowflake’s world-class AI research team, this innovative solution combines the strengths of Mistral’s latest state-of-the-art model, Mistral Large, with Snowflake’s proprietary SQL-generation model. This unique synergy pushes the boundaries of data exploration, empowering you with an unparalleled AI assistant.

Here’s what sets Copilot apart:

Snowflake understands the complexities of writing SQL better than anyone. Processing more than 5 billion queries daily on our platform (as of April 30, 2024) [1] gives us unparalleled insights into the most complex data challenges. This vast amount of data fuels the development of Copilot, surpassing typical large language models. Not only do we have a unique vantage point into the challenges faced by data analysts, we also possess rich metadata that feeds into Snowflake’s dedicated text-to-SQL model, which Copilot leverages in combination with Mistral’s technology. 

In essence, Copilot acts as an intelligent assistant that understands the context of your data, resulting in more accurate and relevant SQL code suggestions (based on internal testing) that can boost your productivity. The powerful combination between Snowflake and Mistral unlocks new use cases and brings you improved capabilities: 

Effortless text-to-SQL: Simply ask Copilot your data questions in plain English, and it will generate the corresponding SQL queries for you.

This empowers users to write SQL queries faster than ever and analyze data with ease.

Enhanced accuracy: Through continuous refinements to Copilot’s modeling architecture, we’ve enhanced its ability to understand user needs.

This combination of Snowflake’s proprietary SQL-generation model and Mistral Large in Copilot is on the frontier for complex real-world SQL-generation tasks.

Data exploration: You can now ask Copilot in plain English to explore your data.

Ask open-ended questions about the structure of your data, refine your analysis with follow-up inquiries and gain deeper insights — all without writing complex queries.

Intelligent corrections: Copilot empowers you to write cleaner, more efficient SQL queries.

It suggests optimizations, provides explanations, and even recommends fixes for potential issues in your existing queries. This comprehensive functionality streamlines your workflow and enables high-quality data analyses.

Snowflake documentation: Copilot can now answer any questions you have about Snowflake documentation. So whether you’re looking for a specific function, would like to learn more about existing features and capabilities, or want to hone your SQL skills — Copilot has you covered with comprehensive Snowflake knowledge.

Seamless AI within Snowflake

Copilot is a complete solution designed to seamlessly integrate with your existing Snowflake workflow. This eliminates the need for external LLMs, keeping your data analysis entirely within Snowflake’s secure environment. It’s readily available alongside your familiar development interfaces, starting with SQL worksheets. Our vision extends beyond worksheets, and we’re actively working to make Copilot a ubiquitous companion in Snowflake.

Powered by Snowflake Cortex, Copilot leverages advanced technologies, like mixed models and efficient GPU processing, behind the scenes. This aligns perfectly with Snowflake’s core mission: democratizing access to the power of AI and empowering everyone to unlock deeper data insights.

Looking ahead: The future of natural language and data

The field of text-to-SQL is undergoing rapid innovation. Snowflake’s AI research team is at the forefront, exploring the intricacies of building a powerful SQL LLM. We’re committed to pushing the boundaries and constantly investigating and testing new possibilities. Our approach isn’t limited to a single model — we use a combination of models, or even entirely new approaches, to improve Copilot’s SQL-generation quality and functionality.

We believe Snowflake Copilot represents a significant step toward a future where natural language becomes the primary interface for data analysis. 

Ready to experience the future of data analysis with Snowflake Copilot? Visit our docs to learn more, including getting started guides and limitations, or check out the FAQs from our AMA