Snowflake introduced a wide range of new AI and data management features at its Summit 2026 event, highlighting its push to help companies move beyond AI testing and into real-world deployment.

The latest updates are aimed at making AI easier to build, manage, and use across organizations. They also focus on helping businesses work with data spread across different systems without the need to copy or move it.

Among the major announcements was an expanded version of Snowflake CoCo, the company’s AI coding assistant. Previously known as Cortex Code, the tool can now be used across desktop and mobile devices, Slack, VS Code, Claude Code, and Microsoft Excel. Snowflake says developers can use natural language prompts to automate tasks, create applications, and simplify development workflows.

The company also launched Datastream, a managed service for Apache Kafka that brings live data streams into Snowflake. The service is designed to help businesses build AI applications and autonomous agents that can respond to constantly changing information.

For everyday business users, Snowflake added several new capabilities to CoWork, its AI assistant formerly known as Snowflake Intelligence. Features such as Cortex Sense, Deep Research, Artifacts, User Skills, and personalization are intended to make it easier for employees to find information and work with company data.

Another addition is Cortex Training, a managed service that allows organizations to customize and train foundation models directly within Snowflake. By keeping the process inside the platform, companies can develop AI models without managing separate infrastructure.

Snowflake also announced new governance features through Horizon Catalog. A new capability called Horizon Context gives employees, applications, and AI agents access to the same business definitions and data context, helping maintain consistency across the organization.

The company introduced Agent Identity and new Trust Center enhancements as part of its security updates. These features are designed to provide organizations with better oversight of AI systems and greater control over autonomous agents.

To help manage workloads more efficiently, Snowflake unveiled Adaptive Compute, which automatically adjusts computing resources based on demand. The company said the feature can improve performance while reducing the need for manual tuning.

Support for open data architectures was expanded as well. Snowflake added new Apache Iceberg capabilities, including Iceberg v3 support, managed storage for Iceberg tables, external engine access controls, and the Iceberg REST Scan Plan API.

According to the company, these updates allow organizations to access and govern data across Snowflake environments, external data lakes, and other open systems without duplicating datasets.

Speaking about the announcements, Vijayant Rai, Managing Director for India at Snowflake, said the next phase of enterprise AI will depend on how effectively organizations connect trusted data, intelligence, and business actions. He said the new capabilities are intended to give businesses a secure foundation for building and scaling AI while enabling people and AI agents to work together using shared business context.

Snowflake also outlined future plans for deeper AI integration. Upcoming features will allow users to interact with enterprise data through natural language, while capabilities such as Automatic Data Agents and Agent Sharing are expected to help organizations create and distribute AI agents with governance controls built in.

Also Read:

Snowflake Launches SnowWork to Automate Multi-Step Business Tasks with AI