Snowflake delivers slew of AI tools, introduces new ones

Adding AI

Snowflake was slower than some of its competitors to add AI development capabilities after the November 2022 launch of OpenAI's ChatGPT marked significant improvement in generative AI (GenAI) technology and sparked surging interest in building AI applications.

The announcements by Snowflake represent significant updates to the Snowflake AI Data Cloud. These updates will better enable the platform to address the development, management and governance of AI workloads and applications as well as facilitate data democratization.

While rival Databricks quickly constructed a GenAI developer framework and tech giants AWS, Google Cloud and Microsoft invested in tools to help customers build GenAI tools, Snowflake didn't fully commit. That changed after Sridhar Ramaswamy became CEO in February 2024, and Snowflake began to prioritize AI development capabilities.

As it adds new features, Snowflake's aims are to provide users with unrestricted access to data, give developers tools that simplify and accelerate the building agents and other applications, and infuse its own platform with agentic AI, according to Christian Kleinerman, the vendor's executive vice president of product.

"The aspiration for us at Snowflake is to help organizations more easily get value from data through the entire data lifecycle," he said during a virtual news conference on Oct. 29. "Of course, AI plays an important role. If business intelligence democratized access to data 20 to 30 years ago, AI is taking that to the next level."

While AI now plays an important role for Snowflake, most of the AI-related capabilities introduced at its Snowflake Summit user conference in June were in preview or testing. Now, in conjunction with its virtual Snowflake Build conference, the vendor began to deliver some of these AI features to GA.

By doing so, Snowflake is showing its ongoing transition from a data management platform to a broader environment that also supports AI application development, according to Mike Leone, an analyst at Omdia, a division of Informa TechTarget.

"The core value lies in aggressively addressing the primary obstacle to AI success, which is increasingly siloed data and fragmented governance," he said. "The entire Snowflake suite, including … Snowflake Intelligence, is designed to help every employee easily unlock intelligence and drive real [value] from their data and AI investments."

Snowflake Openflow is a multimodal data integration feature designed to streamline access to data to remove technical barriers that enterprises face when developing AI and analytics tools. By automating data integration, irrespective of the data type, Openflow brings an enterprise's data together.

Snowflake Intelligence is similarly aimed at providing users with access to data without limits. However, rather than prepare data for analysis, Snowflake Intelligence is an AI agent that enables the analysis itself. With an enterprise's data unified in its data lakehouse, the agent allows customers to use natural language to discover not just what is happening in their data but explain why.

While Snowflake Openflow and Snowflake Intelligence address access to data, Snowflake Cortex AISQL is designed to simplify and speed AI development by enabling developers to build scalable AI pipelines within Snowflake Dynamic Tables using SQL queries.

Likewise, Cortex Agents streamlines and simplifies AI agent development.

Of the features now GA, Snowflake Intelligence is perhaps the highlight, according to Leone.

"Snowflake Intelligence directly democratizes the power of AI by allowing every employee to securely talk to all their company's data in one place using natural language," he said. "This shifts the business value from data insights to immediate, conversational action."

Aslett, meanwhile, called out Snowflake Openflow, for its ability to integrate data from multiple sources irrespective of its format.

Additional GA features include updates to the Snowflake Horizon Catalog to support data lakehouse functionality, Snowflake Dynamic Tables to create AI-inference pipelines with SQL queries, Workspaces to accelerate development through collaboration, integrations with developer platforms such as Git and DBT Labs, and a tool that enables developers to run Apache Spark code on Snowflake's engine.

Meanwhile, Snowflake unveiled the following features either in preview or testing:

• A managed Model Context Protocol Server to standardize how agents interact with external data sources, including large language models.
• Snowflake Postgres, which includes capabilities acquired with the vendor's purchase of Crunchy Data, to enable access to real-time transactional data.
• Cortex Code, an updated version of Snowflake's AI assistant that enables users to interact with the vendor's entire platform using natural language.
• AI Redact, a feature in Snowflake Cortex AISQL that offers users a way to redact sensitive information from unstructured data.
• An updated version of Data Quality User Experience to simplify monitoring data reliability.

"The combination of Snowflake Postgres and enhanced Horizon Catalog is critical as it finally tears down the major architectural roadblock of separating transactional and analytical data," Leone said. "By supporting transactional, hybrid and analytical workloads natively, the platform eliminates data movement and ensures AI agents are built on live, trusted data."

Meanwhile, from a competitive perspective, the new features further demonstrate Snowflake's commitment to AI under Ramaswamy's leadership, he continued.

"Historically, Snowflake has had to fight to catch up in the realm of advanced machine learning and AI workloads that rely on open source and deeply customizable pipelines," Leone said. "New capabilities like Cortex Agents and Snowflake Postgres are clearly targeted moves to close this gap by enabling the development of AI agents on live, trusted data."

Looking ahead

Just as Snowflake on Tuesday made previously introduced features GA while also unveiling new ones, the vendor plans to continue creating new features to stay current with AI innovations while simultaneously advancing tools already GA or in preview, according to Kleinerman.

"AI innovation is happening at an industry level, and we have at this point proven that we are in tune with priorities, trends and opportunities," he said. "We will continue to seize those and deliver on those for our customers. Fast pace of innovation is front and center for us."

While focusing on AI development makes strategic sense, Aslett cautioned that not all customers will be keen to adopt these agentic AI tools. Unlike traditional business intelligence platforms and other enterprise applications with time-tested governance frameworks, it's not yet clear what risks come when tools, such as Snowflake Intelligence, expose data to autonomous agents and non-technical employees.

"Snowflake Intelligence is an interesting new addition that lowers the barriers to gaining insight from data," Aslett said. "Other database providers are delivering similar capabilities to democratize access to data. But it remains to be seen how comfortable enterprises will be with the concept of enabling employees and agents to directly query their analytic database."

Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.