Snowflake’s Cortex AISQL Boosts SQL Performance Up to 70%, Saves Up to 60% Cost

US-based AI data and cloud company Snowflake has announced SnowConvert AI and Cortex AI SQL, further expanding its suite of products and features on the Cortex AI platform. 

SnowConvert AI is an “agentic automation solution” that helps enterprises migrate from legacy platforms to Snowflake. Once their data lands on Snowflake, enterprises can use Cortex AISQL, which allows teams to extract insights across multi-model data and build flexible pipelines using SQL and AI. 

Models from Anthropic, Meta, Mistral, OpenAI, and others power Cortex AISQL. The company said that with Cortex AISQL performance optimisation, enterprises gain performance improvements ranging from 30% to 70%, depending on datasets, with up to 60% cost savings when filtering or joining data across thousands of records.

“Every organisation recognises the potential of AI. But too often, harnessing AI means overcoming complex infrastructure, performance limitations, high costs, and relying on engineers to build custom pipelines,” Carl Perry, head of analytics at Snowflake, said. 

"We’re removing those barriers, whether it’s enabling anyone to analyse and act on all their data with Cortex AISQL or accelerating migrations off legacy systems through SnowConvert AI,” he added. 

Besides, with SnowConvert AI, the company says it can help enterprises complete code conversion and testing phases two to three times faster. “SnowConvert AI goes beyond just database migrations, allowing customers to migrate their entire data ecosystem, from BI tools to ETL tools,” said the company. 

‘2025 is Where We Expect a Lot of Value from AI Systems in Production’

Besides SnowConvert AI and Cortex AISQL, the company announced a slew of updates to its AI platforms, services, and products. 

A few weeks ago, AIM contacted Vijayant Rai, managing director at Snowflake India, to understand the current state of agentic AI in enterprises and the way forward. 

“In 2024, we started seeing [AI] systems going into production…and then in this year, we expect to see a lot of value in terms of these systems going in production,” Rai said. 

Rai added that over the last few months, companies have started to see how AI can be pivotal in changing the way they handle various operations. Thus, many companies with legacy systems started scrambling to try and get their data estates in place to apply AI on top of it, or experimenting with AI straight away. 

“For example, we have a large insurance customer here, and they have a mountain load of PDFs based on court documents. And they use simple search features on Cortex AI to figure out patterns, say for fraud,” Rai said. 

Besides, several Snowflake customers have revealed tangible benefits from Cortex AI. For example, Alberta Health Services is using Cortex AI to enable ER doctors to automate note-taking, which in its proof-of-concept phase has allowed them to treat 10% to 15% more patients per hour.

In another instance, Advisor360°, a wealth manager platform, successfully built an automated customer satisfaction pipeline using Cortex AI in just two days. This solution is estimated to be 1/25th the cost of alternative methods and saves their senior data scientists at least a day's work each month.

“This technology is really good at going through volumes that a human can’t do, and that is something it will excel in. It will find patterns, create commonalities and so on,” Rai said. However, he also reiterated the role of human experts while managing and designing these systems. 

He further noted that domain experts will still be needed to make decisions based on AI results while also ensuring accuracy, security, and data governance. “These three areas, we believe, are still going to be pivotal as people design on top of what we already have,” he added.