Snowflake is launching a series of new features that make it easier for companies to build and deploy AI applications. The data company is introducing Agentic Products on the Snowflake Marketplace and improving its capabilities for collaborating with external data sources. The goal is to remove commercial and technical barriers to AI adoption.
Companies are facing the challenge of turning AI experiments into real business value. Snowflake aims to accelerate this transition by making it easier to share, find, and utilize data for AI applications. The new features focus primarily on connecting internal and external data sources for agentic AI systems.
AI agents directly from the marketplace
The Snowflake Marketplace will support Agentic Products. These are applications that use agentic AI and can be implemented directly on the platform. Data engineers and data scientists no longer need to copy or move their data to benefit from agentic AI.
Providers can develop Agentic Products with pre-trained machine learning models or fine-tuned LLMs that integrate with Snowflake Intelligence. These products run in the customer’s own Snowflake environment, which improves security. This allows teams to integrate third-party agentic experiences into their applications more quickly.
In previous Cortex updates, Snowflake has already focused on lowering the barrier to AI development. With the new marketplace features, the company is taking another step forward in democratizing AI technology.
External data without hassle
Cortex Knowledge Extensions enable data from external partners such as The Associated Press, USA TODAY Network, and Stack Overflow to be used directly in AI assistants. The technology respects intellectual property and commercial terms while ensuring proper attribution.
In addition, organizations can share Semantic Models via the marketplace. This eliminates the time and effort normally required to create semantic models for shared data. Because providers have already created semantic models for their datasets, teams can immediately “talk” to data from suppliers such as CARTO, Cotality, and CB Insights.
These steps make it easier to provide AI assistants and agentic systems with external structured or unstructured data. Information that would normally be behind paywalls is now more accessible to AI applications.
More flexible purchasing and payment models
The Marketplace Capacity Drawdown Program is being expanded. Customers can now use their Snowflake credits to purchase solutions from integrated AI Data Cloud partners dbt, Immuta, and Monte Carlo. This eliminates the need for additional budget approvals and speeds up implementation.
With Offers on Snowflake Marketplace, customers can also negotiate customized terms and pricing for partner products. This provides more partners with the opportunity to offer their solutions on the marketplace, while enhancing ease of use and speeding up the deal-closing process.
Snowflake Native App providers gain more options in monetization models. The commitment + usage model simplifies price negotiations by aligning directly with Snowflake’s pricing model.
Better security and app management
Updates to the Snowflake Native App Framework focus on improved security and interoperability. Providers can now deploy multiple versions of an app simultaneously, enabling seamless rollouts and A/B testing.
New features such as custom metric issuance from app code and session debug mode provide greater observability and simplify troubleshooting. The permissioning system is also getting an upgrade, with automated privilege grants and role-based access controls.
The metrics tab in Snowflake Trail shows CPU/GPU and memory usage of apps, along with important app events such as service fails and restarts. This makes it easier to trace and resolve performance issues.
Privacy-conscious collaboration
Snowflake Data Clean Rooms are gaining new advertising measurement capabilities. The technology integrates media publishers, advertisers, and measurement partners into a neutral environment where they can securely share data and perform customized measurements.
This approach eliminates the need for rigid, one-size-fits-all solutions. Participants gain flexibility to manage ad logs, define custom measurement logic, and gain consistent real-time insights.
Snowflake Horizon Catalog now makes it easier to extract value from sensitive data. Improvements to Horizon Universal Search accelerate the process of finding data, apps, and AI products. Third-party data discovery will also be available soon.
Data managers can now enable secure discovery of sensitive data through a request-for-access workflow within the Snowflake Internal Marketplace. Teams without the appropriate privileges can easily request access, encouraging efficient use of data assets.
Tip: Snowflake automates resource management with Adaptive Compute