Unlock AI's Full Potential: Secure Your Data NOW!

Today, our demo will focus on fictional business Snow Music, which manages live tour operations and high-traffic fan application. As Snow Music prepares for the global tour, tour operations staff across multiple venues are using our Cortex agents to prepare for the show. Our challenge is giving staff access to live fan data without exposing sensitive information like emails, addresses, and payment methods. Configuring unique

guardrails for each role and venue usually takes months. In this demo, I'll show you how Snowflake provides consistent cross-cloud security for your AI agents and data estate, building comprehensive guardrails in minutes.

To see how we secure this agent, I'll start in a Snowflake worksheet to walk you through our pre-configured setup with purpose-built AI controls designed specifically to secure access for your agents. Using data movement policy and agent identity, admins define unique guardrails for autonomous agents. Our policy targets tables automatically tagged with sensitive columns and checks if data has been copied out. Let's say I'm a frontline tour ops

employee. I'm going to ask Co-worker for a list of VIPs with their contact information. Co-worker immediately responds but does not reveal sensitive information. At runtime, agent identity recognized the agent was accessing data on behalf of my restricted role and instantly enforced our masking policies. The employee gets exactly what they need to do their job and nothing they shouldn't. I'll now ask the agent to export our VIP

fan records to share with our brand partners. The agent attempts to execute the stored procedure to copy the data to an external location. This instantly invokes our platform layer data movement policy. This policy detects that the underlying tables carry PII tags and immediately blocks the agentic exfiltration in real time. The agent reports that the export has been blocked and logged. Zero data leaves our secure perimeter. Now, let's

switch to the Snowflake Trust Center, where Snowflake delivers built-in trust and visibility for your AI foundation. Here, the admin sees all violations and detections. With a single click, we get full audit visibility into exactly who prompted the agent and what data was targeted, confirming our data movement policy successfully intercepted the threat, so zero data left our perimeter. Next,

let's look at the AI security tab to monitor our AI security posture in real time. Here, we can see that our out-of-the-box AI scanners are all enabled and actively monitoring for any misconfigurations. This is the full picture of enterprise-grade AI security. Snowflake provides the comprehensive

controls required for regulated businesses to stay resilient in evolving threat landscape. In this demo, we showed how to secure data estate and the agent layer simultaneously, ensuring that threats are automatically neutralized before execution. With that, Snowflake AI security allows you to safely deploy autonomous workflows across an entire organization in minutes instead of months, ensuring data and agent integrity while saving security teams weeks of complex compliance work.