Snowflake opens chat-driven access to enterprise and third-party data

Cortex-powered front end for easier access to insights across multiple sources

Snowflake is set to preview a new platform it claims will help organizations build chatbots that can serve up data from its own analytics systems and those external to the cloud data platform vendor.

Dubbed Snowflake Intelligence, it is built on the vendor's ML and governance technologies and relies on a chat API to access data sources outside Snowflake's platform.

The vendor says developers could connect the chat front end with third-party tools, such as sales transaction databases, knowledge bases like Microsoft SharePoint, and productivity tools such as Slack, Salesforce, and Google Workspace, all within a single Snowflake governance layer.

The point is it is intended to make it easier for developers to build tools that help users "take data-driven actions" and other things consultants like to talk about.

Snowflake Intelligence uses the vendor's managed AI service, Cortex. It uses Cortex Search to run queries on unstructured data, and Cortex Analyst – currently in public preview – to query structured data.

Snowflake Intelligence is also set to come with Snowflake Horizon Catalog, allowing developers to find and use data in open table formats such as Apache Iceberg.

Within Cortex, Snowflake is introducing a new chat API, which it says will help the Cortex Analyst product create conversational experiences with structured data and enable the Cortex Search capability to build conversational experiences with documents.

Cortex AI also offers access to external large language models via LLM functions or REST APIs. "We offer a broad selection of models in various sizes, context window lengths, and language support. Recent additions include the multilingual embedding model from Voyage, the Llama 3.1 and 3.2 models from Meta, and the Jamba Instruct model from AI21," the company said in a blog post.

Christian Kleinerman, EVP of Product at Snowflake, told media that the move is part of the company's efforts to provide developers with a unified experience for working with both structured and unstructured data.

"This higher-level API makes it simpler to create conversational experiences on top of all types of data," he said. "In the past, we had pursued independent paths: Analyst for structured; Search for unstructured.

"We've seen that customers and partners are trying to build a single experience across both types of data ... and sometimes multiple types of data. Cortex Chat API bridges that gap and adds the ability to make API calls, the building blocks required to build agents or agent experiences."

He said the API is designed for individuals or organizations aiming to build applications that enable chat-based data interaction and actionable insights. ®