Snowflake announced updates to Snowflake Intelligence and Cortex Code on April 21, framing its data platform as the “control plane for the agentic enterprise,” according to Snowflake’s press release. The company reported that more than 9,100 customers now use its AI products weekly.
The pitch: agents should run where the data already lives, inside governed infrastructure, rather than exporting data to external agent frameworks.
Snowflake Intelligence as Personal Work Agent
Snowflake Intelligence is evolving from a query tool into a personal work agent that adapts to individual user behavior over time, according to SiliconANGLE. New capabilities entering general availability include Skills, which let users describe multi-step workflows in natural language (preparing presentations, running analysis, sending follow-ups) and have the agent execute them automatically.
New Model Context Protocol connectors link Intelligence directly to Gmail, Google Calendar, Google Docs, Jira, Salesforce, and Slack. An iOS mobile app entering public preview will let users query data and trigger workflows from their phones. A “deep research” feature generates multi-step, cited reports, while “artifacts” let users save and share analyses and workflows across teams.
Cortex Code Goes Cross-Platform
Cortex Code, Snowflake’s AI coding agent for data workflows, has seen adoption from more than 50% of Snowflake’s customer base since launching in November 2025, according to the company. The expansion adds support for external data systems including AWS Glue, Databricks, and PostgreSQL, letting developers build without migrating data into Snowflake first.
On the interoperability side, Cortex Code now integrates through both MCP and the Agent Communication Protocol, enabling connections to external agent frameworks. New integrations include Visual Studio Code and a plugin for Anthropic’s Claude Code. A new Agent SDK supporting Python and TypeScript lets enterprises embed Cortex Code capabilities into their own applications in headless mode.
Browser-based sandboxes for running code without local setup round out the developer experience updates.
The Data-Native Agent Bet
“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,” Baris Gultekin, VP of AI at Snowflake, said in a statement.
The strategy is a direct play against standalone agent platforms. Snowflake’s argument is that agents need governed data access, lineage tracking, and compliance controls from day one, and that data platforms already have this infrastructure. Whether enterprises buy the “control plane” pitch or continue assembling multi-vendor agent stacks will define the next phase of the platform war.