Snowflake renames AI tools amid broader platform push

Snowflake has outlined a broad set of product updates centred on what it calls the "agentic enterprise", introducing new AI assistants, data streaming technology, governance tools and security controls designed to help organisations deploy AI across business operations.

The announcements include new branding for two existing AI products, expanded AI security capabilities, a native streaming service, metadata management tools and additional integrations intended to connect AI agents with enterprise systems.

Christian Kleinerman, Executive Vice President of Product at Snowflake, said the company's strategy is focused on helping organisations combine AI models with enterprise data while maintaining governance and security controls.

Snowflake Intelligence has been renamed Snowflake CoWork and is positioned as a personal work assistant for knowledge workers. Snowflake Cortex Code has been renamed Snowflake CoCo and remains focused on developers and technical users. Kleinerman said customer adoption of the CoCo name influenced the decision to formalise the branding.

Agentic enterprise

Snowflake's vision centres on enabling every employee within an organisation to use AI through access to enterprise systems and data.

The company argues that while AI models continue to evolve, enterprise data remains a more stable asset. As a result, Snowflake is positioning itself as a platform that connects AI models with business data while applying governance, compliance and security controls.

According to the company, the framework for an agentic enterprise combines AI models, enterprise data, SaaS applications and a control layer represented by CoWork and CoCo. These tools are designed to coordinate workflows and interactions between users, data and AI systems.

Snowflake reported that it now serves more than 13,900 customers, including more than 810 Forbes Global 2000 organisations. The company said more than 13,600 accounts use Snowflake AI solutions weekly. It also reported that CoWork accounts have doubled quarter on quarter and that CoCo has reached 7,100 accounts.

Datastream

One of the major platform additions is Snowflake Datastream, a native streaming service integrated directly into Snowflake.

The service is designed to provide drop-in Kafka compatibility for existing Apache Kafka applications and clients. It enables organisations to stream data into Snowflake-managed infrastructure while supporting real-time analytics alongside streaming workloads. Topics can be materialised directly into Snowflake tables, allowing data ingestion and analysis to operate together.

Kleinerman said customers have frequently requested a Kafka-style streaming service that removes operational complexity and infrastructure management.

"The main problem is customers are still dealing with too many moving parts," said Kleinerman, Executive Vice President of Product, Snowflake.

"We have heard consistently that most Kafka-based solutions have a decent amount of fragility, a lot of infrastructure, many aspects that can fail, a lot of management of that infrastructure on part of the customer."

He added that low-latency performance was one of the key technical challenges addressed during development.

Data context

Snowflake also introduced Horizon Context, an extension of its Horizon Catalog designed to improve metadata and semantic understanding across enterprise data assets.

The technology incorporates capabilities acquired through Snowflake's purchase of SelectStar. It can extract metadata from business intelligence platforms, databases and data transformation tools including PostgreSQL, SQL Server, Tableau, Power BI and dbt.

The goal is to provide richer business context and relationships around enterprise data so that AI systems can generate more accurate responses and insights. Horizon Context also extends support for semantic views, which model business relationships and meanings within datasets.

Snowflake also introduced Cortex Sense, which generates additional contextual information from both customer data and patterns of data usage. The company said this capability will improve the quality of outputs produced by CoWork and CoCo with less manual configuration.

AI security

Security formed another major focus of the announcements.

Snowflake is launching Trust Center: AI Security Package, which scans for risks and vulnerabilities associated with AI agents and AI-driven workloads. The package is designed to identify exposures and alert organisations through the company's Trust Center capabilities.

New data movement policies allow organisations to restrict actions performed by AI agents, even when the underlying user account has broader permissions. For example, companies can block agents from exporting data to external cloud storage destinations.

The company is also introducing expanded ransomware protection capabilities under its Trust Center framework. One feature, called multi-party approvals, requires two administrators to approve sensitive operations. Snowflake said this is intended to reduce risks associated with insider threats or unintended actions by AI agents.

Snowflake is also adding agent identity capabilities. Under the approach, AI agents inherit permissions from their creators but can be further restricted through dedicated agent-specific policies.

Interoperability focus

Snowflake announced several interoperability updates aimed at reducing dependency on proprietary architectures.

The company has added support for Iceberg V3 and introduced interoperability between Horizon Catalog and Apache Iceberg REST APIs. Snowflake can also provide storage for Apache Iceberg tables.

Kleinerman said interoperability remains a strategic priority.

"We are not only adopting Apache Iceberg and Apache Polaris as core standards. We have a number of members of the steering committee guiding the effort of Iceberg so that it is truly interoperable," said Kleinerman.

Personal assistants

The company is expanding CoWork's role as a personal work assistant.

Each user can have a personalised work agent with dedicated skills, connectors and memory. Snowflake is also introducing scheduled automations and new dashboard capabilities that support collaboration, sharing and recurring analysis workflows.

Additional integrations are being added through MCP connectors. Snowflake also cited its acquisition of Natoma, which provides an MCP gateway designed to manage secure connectivity between agents and external systems.

For developers, Snowflake is expanding CoCo through CoCo Desktop, a VS Code Extension, a Claude Code Plugin, a CoCo Slackbot and a Skill Catalog.

"We want companies to be able to leverage AI capabilities with governance," said Kleinerman. "We're incredibly excited about our innovation. And of course, we are committed to continue to innovate to help our customers achieve better business results."