Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders. Subscribe Now
A day after launching an integration for productivity platform Coda, data cloud company Snowflake has announced the launch of its Copilot – an intelligent SQL (structured query language) queries assistant – in public preview.
First announced at the company’s Snowday event last year, Snowflake Copilot taps the power of Snowflake’s proprietary text-to-SQL model and Mistral’s recently launched Mistral Large large language model (LLM) to generate relevant SQL queries and help users understand and explore their data. Snowflake invested an undisclosed sum in the Paris-based startup last month to bring its entire family of models to its Cortex service for LLM app development.
The move marks another notable step from the data cloud giant to tap the power of AI to simplify how enterprises work with their data assets — a strategy that has taken shape aggressively ever since Sridhar Ramaswamy, who came in from the acquisition of Neeva AI, took over as CEO.
What to expect from Snowflake Copilot?
Snowflake sits at the center of the data revolution, allowing enterprises to analyze data assets and pull out relevant insights for decision-making. The approach has made the company what it is, but extracting insights has largely revolved around writing complex SQL – which takes time and is not suitable for every enterprise user.
AI Scaling Hits Its Limits
Power caps, rising token costs, and inference delays are reshaping enterprise AI. Join our exclusive salon to discover how top teams are:
- Turning energy into a strategic advantage
- Architecting efficient inference for real throughput gains
- Unlocking competitive ROI with sustainable AI systems
Secure your spot to stay ahead: https://bit.ly/4mwGngO
With the new Copilot, which is rolling in public preview for AWS accounts in the U.S., Snowflake is solving this challenge. According to multiple demos shared by the company, the assistant will sit quietly within SQL worksheets and give users a conversational interface to generate SQL queries in natural language. Users have to just pull up the Copilot from the “Ask Copilot” button and describe what’s needed in English. Soon after this, the bot will understand the question, process it and produce a ready-to-use SQL code to run and achieve the desired result.
Snowflake says customers can use Copilot’s generation capabilities for a variety of tasks, including extracting data from multiple tables for analysis as well as correction of existing queries. And, if users have no idea where to start, they can also engage in a back-and-forth conversation with the assistant to understand the structure of the dataset and what questions should be asked to get insights. The Copilot understands the context of the data and accompanies every generated query with a detailed response covering all aspects that led to the answer, including the tables joined.
To build this experience, Snowflake tapped into Cortex, its own service for LLM app development, and used vast amounts of SQL query data and metadata to power its proprietary text-to-SQL model and Mistral Large.
“Processing over 4 billion queries running daily on our platform 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 that Copilot leverages in combination with Mistral’s technology,” Snowflake AI’s senior product manager Pieter Verhoeven and Principal AI engineer Yusuf Ozuysal wrote in a joint blog post.
Expansion in pipeline
With the public preview of Copilot, Snowflake hopes to gather user feedback to refine the solution and make it ready for general availability — the timeline of which remains unclear at this stage. The assistant currently remains restricted to SQL worksheets, but the company has hinted at the next step of evolution, where it will be expanded to other parts of the product, serving as a “ubiquitous companion” of sorts for users.
While it remains to be seen what these areas would be, the result is not hard to expect: easy access to insights from Snowflake for faster decision-making. This could be a game-changer for the company, but it is important to note that Snowflake is not the only one experimenting with natural language querying. Ever since LLMs came to the scene, many data players have explored simplifying their product experiences with generative AI. In terms of querying with LLMs, Dremio and Kinetica have also launched similar capabilities.
