Snowflake has announced updates to Snowflake Intelligence and Cortex Code, positioning the tools as part of a broader push to support “agentic” AI use cases, where systems move beyond answering questions to taking actions across enterprise workflows.
The company said the enhancements are designed to let organisations connect more data sources, enterprise systems and AI models with data held in Snowflake, within a unified experience. Snowflake is framing the combined capability as a control plane for what it calls the “agentic enterprise”.
Snowflake Intelligence is being presented as a work agent for business users, intended to adapt over time by learning individual preferences and workflows. Snowflake said the tool aims to deliver more relevant results, automate tasks, and provide insights grounded in governed enterprise data.
Among the announced additions are “Skills”, described as workflow automations created in natural language and slated to become generally available soon; Model Context Protocol (MCP) connectors (also “generally available soon”) intended to integrate with tools including Gmail, Google Calendar, Google Docs, Jira, Salesforce and Slack; and an iOS mobile app, expected to enter public preview soon. Snowflake also flagged a “deep research” capability (public preview soon) for multi-step reports with citations, and “Artifacts” (generally available soon) to save and share analyses and workflows.
Snowflake said the Snowflake Intelligence updates were informed by customer feedback and a recent research preview launch of Project SnowWork.
For developers and data teams, Snowflake said Cortex Code is expanding to support building and operationalising AI across “increasingly complex, multi-system data environments”. The company said Cortex Code, launched in November 2025, is being used by more than half of its customers.
New capabilities outlined include support for additional external data systems such as AWS Glue, Databricks and Postgres; integrations via MCP and Agent Communication Protocol (ACP); a VS Code extension (private preview) and a Claude Code plugin; and an Agent Software Development Kit supporting Python and TypeScript. Snowflake also described “Cloud Agents” (private preview) for running code and workflows in a managed browser-based environment, along with “Plan Mode” for previewing workflows and “Snap & Ask” to interact with data artefacts such as charts and tables.
“AI is changing how every company operates, and the platforms that win will make it easy to put AI into practice with the right data and guardrails,” said Baris Gultekin, VP of AI at Snowflake. “Snowflake gives customers one place to bring their data together, connect the systems they rely on, and turn AI into something that actually helps teams get work done.”
Snowflake said more than 9,100 customers use its AI products weekly. In the announcement, the company cited examples of organisations it said are moving AI into production on its platform, including Capita, Logitech, Telenav, United Rentals and Wolfspeed, and included partner commentary from Accenture.
The updates come as enterprises weigh how to operationalise AI agents while maintaining governance and security controls over data access, tool integrations and automated actions. For security and risk teams, features such as connectors into common collaboration and business systems, combined with AI-driven task execution, may increase the importance of oversight of identity, permissions, audit trails, and how governed data is exposed to models and agents.
