AI innovation continues to transform applications and experiences across industries and enterprises. Companies are increasingly focused on driving measurable outcomes for their end users. Achieving those outcomes requires AI that is scalable, secure and deeply connected to enterprise data.
At Snowflake, we are committed to helping customers turn AI and ML ambition into real-world impact. That means putting developers in the driver’s seat with tools that make it easier to build reliable agents, accelerate AI/ML workloads into production and govern those workloads confidently as they scale.
Our latest product innovations give customers the ability to build reliable, enterprise-grade applications with Snowflake. The result is faster execution, easier operations and AI tools that enterprises can confidently rely on in production.
Snowflake Intelligence brings together a set of capabilities designed to help business users get value from AI quickly, securely and independently. These updates focus on three core needs:
Snowflake Intelligence is designed to deliver trusted insights wherever work happens. Its natural language interface enables every employee to ask questions, uncover the “why” behind the “what” and take timely, data-driven action, all within Snowflake’s secure and governed platform.
Together, these capabilities make Snowflake Intelligence a trusted enterprise intelligence agent that delivers insights when and where users need them, supporting timely, data-driven decision-making across the organization.
Artifacts (public preview soon) represent a foundational shift in how Snowflake Intelligence supports business users. Artifacts turn Snowflake Intelligence conversations into saveable, shareable outputs such as charts and tables that preserve visualization, underlying SQL and contextual metadata.
Artifacts are the core units through which enterprise knowledge is captured, shared and acted upon in Snowflake Intelligence. Users can save artifacts to avoid recreating analyses, share live references securely with teammates and explore follow-up questions in context. Artifacts allow users to return to what they’ve built, share it with others and collaborate directly on trusted enterprise data.
More broadly, Artifacts are foundational to Snowflake Intelligence’s ability to deliver business insights to end-users. Instead of Snowflake Intelligence being used primarily for ad-hoc or follow-up questions, artifacts help make it the starting point for driving the business. With artifacts, we’re making Snowflake Intelligence the destination for consistent and trustworthy decision-making from a single source across the organization.
Snowflake Intelligence will soon be available as an iOS mobile app (public preview soon), offering an improved native mobile experience. Mobile access ensures leaders and business users can stay connected to their enterprise knowledge throughout the day, whether they are reviewing key metrics, monitoring trends or following up on critical questions as decisions unfold.
For a secure and easy-to-use experience, Snowflake Intelligence will support FaceID-based session renewal on the mobile app (public preview soon). Users will be able to authenticate using FaceID, and tokens will refresh automatically in the background. Refresh tokens remain protected, device-bound and regularly rotated, enabling enterprise-grade security controls while delivering a smooth, consumer-style mobile experience.
Snowflake Intelligence now supports direct user login, so business users can sign in and start asking questions without needing to know Snowflake or navigate Snowsight.
For customers who want tighter control, Snowflake Intelligence-only users give business users access to Snowflake Intelligence and nothing else. They cannot access Snowsight, SQL interfaces, or other data tools. This keeps business users in an interface built for them, while helping organizations govern usage, control costs, and implement all existing security policies automatically.
Snowflake Intelligence also supports identity provider redirect, so users authenticating through a configured IdP (e.g. Okta or Entra ID) have a simplified Snowflake Intelligence login experience. Together, these capabilities make it easy to expand access across the organization while keeping governance controls centralized.
AI agents are now central to business workflows. Enterprises need a reliable and trustworthy stack to deliver consistent and accurate experiences in a governed environment that can scale across teams and applications. We are excited to share major innovations on Snowflake that help customers build and scale production-grade agents with confidence.
Cortex Code, now generally available, supports this journey by enabling every builder — from seasoned engineers to nontechnical teams — to build and optimize agents using natural language interactions. It helps teams easily generate synthetic data, create and debug semantic views and quickly build and debug agent behavior, speeding time to production on the Snowflake AI Data Cloud.
With Semantic View Autopilot (generally available soon), teams can automatically create and deploy production-ready semantic views. By learning from query history, Semantic View Autopilot streamlines modeling workflows and helps organizations onboard new use cases more quickly while delivering consistent insights across teams.
To extend adoption of agents across the organization, Cortex Agent Sharing (generally available soon) makes it easy to discover, reuse and operationalize agents built by internal teams or partners. This enables organizations to standardize agent capabilities, avoid duplication of effort and scale proven agents across teams rather than rebuilding them for each use case. Teams can access offerings from the Snowflake Marketplace and leverage partner-built agents to accelerate time to value.
With Agent Evaluations (generally available soon), customers gain deeper visibility into how agents reason, select tools and generate responses to refine agent behavior and continuously improve accuracy as agents evolve. This transparency helps teams instill confidence in agent quality by easily validating accuracy and logical consistency to ensure they are ready for production workloads. By providing full visibility into an agent’s “thought process,” Agent Evaluations reduce guesswork in debugging, allowing teams to instantly pinpoint and fix errors or performance bottlenecks. Finally, by validating answers, logic and tool usage, organizations can confidently advance agents from early experimentation into production-ready systems that teams trust.
Snowflake Intelligence supports the Model Context Protocol (MCP) to simplify integration with third-party tools and services. We launched a Snowflake managed MCP server in October 2025 and are now following up with Snowflake MCP Client (generally available soon), so customers can connect with external data sources with ease and trust.
With Snowflake MCP Client, account admins can register prebuilt or custom MCP servers, such as Atlassian, Salesforce or Workday, and surface these directly into Cortex Agents. Developers can use MCP servers with agents, enabling seamless tool discovery and invocation during orchestration. Snowflake manages authentication, including token handling, and provides observability to keep integrations secure and governed. At launch, Snowflake supports full MCP tool discovery during agent invocation along with monitoring and token management, enabling customers to securely access and act on enterprise data across systems.
In production environments, consistency and accuracy play a critical role in user experience and adoption. Snowflake continues to invest across the agent stack to deliver AI-powered experiences that are faster, more accurate and predictable at scale.
Snowflake is introducing Continuously Learning Agent Memory (public preview soon), a major quality enhancement for enterprise intelligence agents. This capability enables agents to continuously learn from high-quality past responses across users, improving consistency and trust. It also enables agents to remember individual user preferences and facts over time, delivering more personalized Snowflake Intelligence experiences.
By integrating text-to-SQL inline with agent orchestration, Snowflake has also improved accuracy and reduced latency for analytical workflows. Users can access data faster, view LLM planning alongside SQL execution and refine agent behavior across a broad range of workloads.
As AI applications evolve, enterprises need governance capabilities to scale. Snowflake delivers this through agent versioning and integrated operational visibility.
Agent versioning (public preview soon) brings CI/CD support to Snowflake Cortex Agents so customers can build, deploy and iterate agentic workloads with confidence. Developers can snapshot versions, manage changes through Git and promote or roll back deployments safely. In addition, customers can track usage across Snowflake Intelligence and agents through usage views (generally available soon), supporting better operational oversight.
Beyond visibility, Snowflake enables teams to actively control AI costs. AI_COUNT_TOKENS (generally available) helps estimate usage before execution, while AI Functions Incremental Metering View (generally available soon) will provide usage and cost data of running queries enabling teams to enforce limits and trigger actions during runtime. This allows organizations to scale AI in production while maintaining predictable spend and operational control.
Together with versioning and cost tracking, teams can move quickly while maintaining clarity and scaling high-performance applications responsibly.
Traditional machine learning (ML) remains critical in today’s AI landscape and we are thrilled to announce new capabilities for agentic, multimodal and real-time workflows in Snowflake ML.
We are continuing to invest in modern development experiences that increase productivity. The next generation of Snowflake Notebooks (generally available) are now first-class citizens inside Snowflake Workspaces, running in a Jupyter-based environment powered by Snowflake’s Container Runtime. Snowflake Notebooks enable developers to bring existing Jupyter-based notebooks, scripts and model training into Snowflake’s unified platform for advanced model development workflows. The power of Snowflake Notebooks for more powerful development and iteration is further enhanced by the integration with Cortex Code in Snowsight (generally available soon).
Data scientists often spend lengthy cycles on developing and troubleshooting their ML workflows, leading to operational bottlenecks and fewer ML models making their way to production. Now Snowflake is bringing agentic AI to ML workflows with the integration of Cortex Code for ML workflows in Snowflake Notebooks to autonomously iterate, adjust and generate a fully executable ML pipeline from simple natural language prompts.
Real-time ML models can easily be productionized in Snowflake ML with online feature store and online model serving, both in general availability now. Developers can now serve features in under 30 ms and models in under 100 ms to power low-latency, online use cases such as personalized recommendations and fraud detection — no extra infrastructure or complicated configuration required. Additionally, the ability to run large-scale inference with multimodal models from hubs such as Hugging Face is now in public preview. Inference with unstructured data such as images and video unlocks AI use cases such as object detection, visual Q&A and automatic speech recognition on Snowflake without complex pipelines or data movement.
Today’s launches help establish Cortex Agents as the unified foundation for enterprise-grade AI. Semantic View Autopilot helps developers drive higher Cortex Agent accuracy and accelerates the rollout of advanced use cases. The latest Snowflake ML enhancements enable developers to build models that Cortex Agents can leverage to directly deliver ML-based predictions and recommendations to users. And during production, our Evaluations for Cortex Agents ensures agent outputs are both trustworthy and easy to monitor.
With Snowflake, enterprises move from experimentation to production with AI agents and applications trusted by teams, managed by operators and connected directly to business impact.
1. Start creating, saving and sharing artifacts in Snowflake Intelligence to drive collaboration and action.
2. Explore the Cortex Code announcement.
3. Learn more about the latest ML announcements in this blog.
For years, the promise of a truly “self-driving” data platform has been hindered by the manual complexity of data engineering. […]
Guided by the mission of simplifying data workflows for every builder, wherever they operate, Snowflake is unveiling a series of […]
AI innovation continues to transform applications and experiences across industries and enterprises. Companies are increasingly focused on driving measurable outcomes […]
Our services help both the big and small organisations in the identification of security breaches and provide outclass solutions to prevent any potential harm.