At enterprise scale, the hardest problem is not the intelligence itself. The real challenge is context. Agents fail not because they are not smart enough, but because they do not have enough context about your business, your data and your specific situation. Snowflake Intelligence operates where your most important data already lives, so every answer is grounded in what is actually happening in your business, using your semantic models and the meaning your organization has already defined.
Snowflake Intelligence provides a single conversational interface that spans the entire enterprise data estate. The agent automatically determines where to retrieve information: structured data in Snowflake tables, unstructured content such as documents and transcripts or external systems connected through MCP connectors. Users do not need to know how the system is structured or where data lives. They simply ask, and the agent figures out the rest.
Consider what this looks like for a sales leader preparing for a weekly forecast review. Today, that process means opening multiple dashboards, exporting reports and manually flagging at-risk deals. With Snowflake Intelligence, it becomes one single conversation:
“Which deals are most likely to slip this quarter?” The agent analyzes pipeline data, engagement trends and surfaces, deals with declining momentum and explains the contributing factors.
“Draft follow-ups for the top five at-risk accounts.” The agent generates personalized emails using CRM notes, meeting summaries, and account history.
“Post the summary to the sales channel.” The agent sends it directly to Slack (generally available soon).
What previously required coordinating across a CRM, an email client, and an analyst now happens in one conversation, in minutes.
Fig 1: MCP Connectors to Connect Data and Take Action Across Systems
Fig 3: Skills: Automating End-to-End Workflows from Insight to Execution
The ability to connect insights to action depends on trust, and that trust comes from where the intelligence operates.
Snowflake Intelligence runs directly on the platform that already holds enterprise data. It applies the same governance model organizations rely on today, including role-based access controls, row-level policies and data masking. Every response reflects the data the user is authorized to see, and every action is executed within the boundaries defined by administrators. Budget controls (generally available) provide centralized visibility into AI usage and allow teams to manage cost at the individual team or workflow level. Identity provider integration (generally available), including Okta and Microsoft Entra ID via SCIM, allows organizations to provision business users at scale without manually setting up individual Snowflake accounts. Snowflake Intelligence-only users gain access to the intelligence layer without visibility into Snowsight or SQL interfaces, keeping the experience relevant to their role.
This is a meaningful distinction when compared to general-purpose AI tools. Snowflake Intelligence combines governed access to external systems, with direct access to the full enterprise data estate. Every action is executed within defined policies. Every interaction is fully auditable, and that auditability is what allows AI to move from experimentation into production.
Not all questions can be answered with a single query. Some require connecting signals across multiple systems to surface relationships that are not visible in any single dataset.
Deep Research (public preview soon) extends Snowflake Intelligence for exactly these scenarios. It performs multi-step analysis across data silos, synthesizing results into a structured, fully cited report that explains what is happening, why, and what to do next. Where a standard query returns a single answer, Deep Research runs multiple agents simultaneously, scanning structured data, unstructured content, and external context together, to answer the complex “why” questions that typically require days of cross-functional effort. This complements Extended Thinking’s precise, single-turn depth with broader, multi-source context across the full enterprise data estate.
A product team investigating churn asks why a specific customer segment is leaving at a higher rate than expected. Deep Research analyzes usage data, support tickets, feedback, and sales interactions simultaneously, surfacing contributing factors in order of significance and providing recommendations the team can act on immediately.
Fig 4: Deep Research: Multi-Agent, Cross-Source Analysis for Complex ‘Why’ Questions
The Snowflake Intelligence iOS mobile app (public preview) brings the full Snowflake Intelligence experience to any device, so users can act on an insight, approve a recommendation, or check on a standing goal from anywhere. Face ID authentication removes login friction, so a glance is all it takes to get straight to your personal agent and pick up exactly where you left off.
Every morning business users start their day the same way: opening multiple tools, waiting on updated reports, and reaching out to an analyst for a number they needed yesterday so they can take meaningful actions. The tools and data exist. But nothing connects them or helps them make progress. Snowflake Intelligence changes that.
With the latest updates, Snowflake Intelligence is now a personalized work agent for every business user, one that learns how individuals access their data, derive insights, and take action across the tools they already rely on.
Snowflake Intelligence gives business users one place to ask questions across their data and take action.This personalized work agent produces results grounded in business context and helps users gain a shared understanding of their enterprise data.
Business users can directly operate across the systems where work happens through governed integrations: MCP connectors (generally available soon) that can connect directly with Gmail, Google Calendar, Google Docs, Jira, Salesforce and Slack, allowing users to take action without leaving the workflow. The new Snowflake Intelligence iOS mobile app (public preview) and performance improvements to response latency help ensure the experience is responsive and available wherever work happens. Powered by Cortex Agents, Snowflake Intelligence runs on the same platform that already holds your enterprise data and is governed by the same policies that protect it. They can move from experimentation to driving real business outcomes all within a trusted, governed environment.
With the new capabilities, Snowflake Intelligence now represents a shift from read-only insights to real action, the foundation of Snowflake’s broader vision: to become the control plane for the agentic enterprise.
Fig 5: Your Agent, Anywhere: Snowflake Intelligence on Mobile
Scaling AI across an enterprise requires a platform that enforces governance policies consistently, supports continuous improvement, and gives builders the tools they need at production scale.
Cortex Agents provide the foundation. Builders use composable building blocks to define workflows, integrate tools and data sources, and assemble the capabilities that power the end-user experience. The platform supports the full lifecycle from design through deployment and monitoring. Agent Versioning (generally available) and CI/CD workflows allow teams to iterate safely, roll back when needed, and promote changes with the same engineering rigor applied to production software. Agent Evaluations measure accuracy and reliability at each stage, providing clear signals when quality needs to improve. A secure code execution sandbox (public preview soon) supports advanced data transformation, statistical analysis, and content generation inside the agent workflow.
Over 9,100 customers use Snowflake’s AI products on a weekly basis, and that number continues to grow as enterprises move from AI experimentation to real-world deployment.
Snowflake Intelligence builds on the foundation that enterprises already trust. Bringing intelligence directly to that foundation, rather than extracting data into an external system, is what makes AI practical at the scale and trust level enterprises require.
Snowflake Intelligence is where that vision is becoming operational for business users today, the personalized work agent that every business user needs, and the foundation of Snowflake’s control plane for the agentic enterprise.
Forward-looking statements
This article contains forward-looking statements, including about our future product offerings, and are not commitments to deliver any product offerings. Actual results and offerings may differ and are subject to known and unknown risk and uncertainties. See our latest 10-Q for more information.
The distinction between answers and outcomes becomes clearest when applied to the full range of daily work across functions.
Take a finance analyst investigating a budget variance. They ask: “Why did operating expenses increase in the Northeast last quarter?”
The agent traces the variance across cost centers, supplier invoices and time periods. It:
Identifies the specific cost driver
Provides a breakdown by line item
Explains the context in plain language
The analyst then asks the agent to generate a summary for leadership and notify procurement. Both steps complete in seconds. Investigation to communication, done in one workflow.
Insights can be visualized, saved and shared as Artifacts (generally available soon), reusable, interactive outputs that preserve the underlying data, SQL and context. Artifacts make it possible to share findings inside Snowflake Intelligence. One analysis becomes a living, shared resource that teammates can build on and refine together, with governance controls intact.
Fig 2: From Answers to Outcomes: Turning Analysis into Actionable, Shared Workflows
Users can further streamline these workflows through Skills (generally available soon). A Skill turns a repeatable task into a reusable workflow that any user can invoke with a single prompt. Preparing for a customer meeting, looking up consumption data, generating a briefing, can be defined once and triggered with a single request. A weekly executive summary, a pipeline risk report, a follow-up sequence based on meeting transcripts, can be automated and shared across the team.
The same pattern applies to operations. An operations manager asks: “Are there any inventory risks this week?”The agent checks inventory levels, supplier lead times, and logistics timelines. It flags a potential shortage in one product line and explains the root cause. The manager’s next move:
“Escalate to the supplier.” Done.
“Open a Jira ticket for the logistics team.” Done.
Both actions execute within governance boundaries. No manual coordination across systems required.
A single interface for enterprise intelligence and action At enterprise scale, the hardest problem is not the intelligence itself. The […]
Snowflake Intelligence is redefining how business users work with data—turning insights into decisions and actions across the enterprise. Powering this […]
Get hands-on The best way to see what Cortex Code can do is to run it in your own Snowflake […]
Our services help both the big and small organisations in the identification of security breaches and provide outclass solutions to prevent any potential harm.