Snowflake Copilot: New AI-Powered SQL Assistant Explained

Data Cloud company Snowflake released Copilot, a breakthrough structured query language (SQL) assistant – in the public preview mode. This is a high-level technical advancement from Snowflake and may show promising results in the days to come.

So, what are its groundbreaking features? How can it ease the workload for developers? In this article, we are going to explore all these topics in-depth. So, let’s get right into it!

Highlights:

• Snowflake releases Copilot, its breakthrough AI-powered SQL assistant.
• Comes with several features such as Data Exploration, effortless Text-to-SQL and much more.
• Released in public preview for AWS accounts in the United States.

The New AI-powered SQL Assistant

Copilot from Snowflake is here to simplify the text-to-SQL data analysis process. In today’s world of Generative AI, data analysis with complex SQL queries has become a huge problem and hectic task for several developers. This is why they has launched Copilot, in a mission to mitigate this problem.

Snowflake Copilot is a Gen AI-powered SQL Assistant designed to make complex SQL queries accessible through simple natural language prompts. The data revolution revolves around Snowflake, which enables businesses to examine their data assets and extract pertinent information for use in decision-making.

Snowflake Copilot, which was first unveiled at the company’s Snowday event last year, uses Mistral’s recently released Mistral Large large language model (LLM) and their proprietary text-to-SQL model to construct pertinent SQL queries and assist customers in comprehending and analyzing their data.

This special combination expands the possibilities for data analysis and gives you access to an unmatched AI assistant.

The vendor highlighted that Snowflake, which performs over 4 billion queries every day on its Data Cloud platform, has a profound understanding of SQL complexity, which has benefited Copilot’s development. Their text-to-SQL model is further enhanced and fueled by this technique.

“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 the public preview of Snowflake Copilot, a new solution on the bleeding edge of text-to-SQL that simplifies data analysis while maintaining robust governance.”

In many company demos, the assistant is shown sitting quietly inside SQL workbooks, providing users with a natural language conversational interface to create SQL queries.

To use the Copilot, simply click the “Ask Copilot” button, describe the situation in English, and the Copilot will appear. The bot will then comprehend the query, process it, and generate ready-to-run SQL code that will accomplish the intended outcome shortly after.

Looking Into the Features: What makes it unique?

Snowflake is the only one who truly comprehends the difficulties involved in developing SQL. According to them, users can update existing queries and retrieve data from many tables for study among other uses for Copilot’s generating capabilities.

Additionally, if users are at a loss for where to begin, they can converse back and forth with the assistant to grasp the dataset’s structure and what inquiries should be made to gain insights. Copilot’s fusion with Mistral Large puts it on a pedestal of SQL Data Analysis Assistants.

This fusion archives an execution accuracy percentage of 46.4%, according to the data lake vendor, surpassing industry leaders such as GPT-4 (46%) and CodeLlama 34B Instruct (30.8%).

Snowflake Copilot provides the following functionality and use cases in addition to text-to-SQL:

1) Data Exploration

Now, users can ask Copilot to explore their data in simple English. Without creating complicated queries, they can obtain deeper insights, improve their analysis with follow-up questions, and pose open-ended questions regarding the structure of their data.

Imagine automating your data analysis tasks without the need for complex queries. All of this is made possible with Snowflake’s Copilot.

2) Effortless text-to-SQL

Copilot allows users to ask simple questions about data in plain English, and it will create the appropriate SQL queries on their behalf. Users can now write SQL queries more quickly than before and easily analyze data thanks to this.

So even if you are not adept with SQL Queries and data processing, Snowflake Copilot has got your back.

3) Enhanced Accuracy

The modelling architecture for Copilot has been continuously improved, improving its comprehension of user needs. For complex real-world SQL-generating tasks, this combination of Mistral Large in Copilot and Snowflake’s proprietary SQL generation model is cutting edge.

So now you can enjoy a higher level of improved accuracy in comprehension of your data analysis tasks and SQL generation from Copilot’s SQL AI assistant.

4) Snowflake documentation

You can now ask Copilot any questions regarding the documentation for Snowflake. So, whether you’re searching for a particular function, want to learn more about the features and capabilities that are already there, or want to improve your SQL skills, Copilot’s extensive Snowflake knowledge has you covered.

Understanding the full documentation of Snowflake for a long time has been a difficult and hectic task, but Copilot is here to address all your doubts.

5) Intelligent corrections

You can construct more effective, clearer SQL queries with Copilot’s help. It offers explanations, optimization suggestions, and even fixes for possible problems with your current queries. This extensive functionality allows for high-quality data analytics and optimizes your process.

So, Copilot is not only your Data Analysis assistant but also your AI teacher. It will guide and help you with fixes, tips, and solutions in every step of your query-processing journey.

Overall, Copilot functions essentially as an intelligent assistant that comprehends the context of your data to provide more pertinent and accurate SQL code recommendations (derived from internal testing) that can increase your output.

Are there any limitations?

Snowflake detailed some of Copilot’s current limitations, such as the fact that the tool cannot access the data in users’ tables, and that cross-database queries are not supported.

The responses can take a while to process and can take up to an hour to identify newly created databases, schemas, and tables, and it occasionally suggests queries with incorrect SQL syntax or that don’t contain any tables or columns.

The Future of Natural Language and Data Analysis

Although data is essential to modern businesses, obtaining its full insights frequently necessitates the use of sophisticated SQL queries. It can take a lot of time to write and manage these questions. Snowflake is addressing this issue with the new Copilot, which is now available for public preview for AWS accounts in the United States.

Its vast features and capabilities have tremendous potential to simplify data analysis and processing tasks by removing the need for hard coding with SQL queries. Copilot uses cutting-edge technology including mixed models and effective GPU processing in the background, all powered by Snowflake Cortex.

This is in complete harmony with the company’s primary goal of enabling everyone to gain deeper data insights and democratizing access to AI capability.

Innovation in the text-to-SQL space is happening quickly. The AI research team at Snowflake is leading the way in examining the complexities of creating a potent SQL LLM. It makes use of any combination of models, or even completely novel techniques, to raise the usefulness and quality of Copilot’s SQL generation.

Also in the coding space, GitHub recently launched Code Scanning Autofix, which can be used to use AI to find and fix code vulnerabilities.

Conclusion

Copilot is here to disrupt the world of natural language processing and data analysis forever. Not only does it leverage the developer community with high-level query processing and task automation processes but it also acts as a guide to those interested in learning and exploring Data Queries with SQL.