Snowflake Copilot: AI Assistant Powered by Mistral Large Now Available for Public Preview

A day after launching an integration for the productivity platform Coda, data cloud company Snowflake has unveiled its Copilot – a public preview of an intelligent SQL (Structured Query Language) query assistant.

Initially announced at Snowflake’s Snowday event last year, Copilot utilizes Snowflake’s proprietary text-to-SQL model alongside Mistral’s Large Language Model (LLM) to generate relevant SQL queries and aid users in exploring their data. Snowflake has recently invested in Mistral to integrate its suite of models into its Cortex service, enhancing capabilities for LLM app development.

As Snowflake continues to leverage AI to improve data management for enterprises, this launch exemplifies a strategic shift under CEO Sridhar Ramaswamy, who joined the company via the acquisition of Neeva AI.

What to Expect from Snowflake Copilot?

Snowflake is transforming the data landscape by enabling enterprises to analyze their data assets and derive valuable insights. Traditionally, extracting these insights involved complex SQL writing, a time-consuming process that isn't user-friendly for everyone.

The new Copilot, currently in public preview for AWS accounts in the U.S., addresses this issue by providing a conversational interface within SQL worksheets. Users simply click the “Ask Copilot” button to ask questions in plain English, and the assistant swiftly generates SQL code to achieve the desired results.

Copilot can assist with various tasks, such as extracting data from multiple tables and correcting existing queries. Users unfamiliar with SQL can engage in a dialogue with Copilot to understand dataset structures and formulate insightful questions. Each generated query is accompanied by a detailed explanation, highlighting the joined tables and thought processes involved.

Leveraging its Cortex service, Snowflake harnessed extensive SQL query data and metadata to enhance its proprietary text-to-SQL model and Mistral’s technology.

“Processing over 4 billion queries daily on our platform allows us unmatched insights into complex data challenges. This wealth of data drives the development of Copilot, giving us a unique perspective on the obstacles data analysts face,” said Pieter Verhoeven, Senior Product Manager at Snowflake AI, and Yusuf Ozuysal, Principal AI Engineer.

Future Expansion Plans

With the public preview of Copilot, Snowflake aims to gather user feedback to refine the tool before general availability, though the timeline is still uncertain. While currently limited to SQL worksheets, the company suggests plans for expansion into additional areas of its product, positioning Copilot as an essential resource for users.

As other data players like Dremio and Kinetica also explore natural language querying, the advancements from Snowflake could enhance accessibility to insights, ultimately speeding up decision-making processes.