Autonomous AI agents can write code, execute commands, and make decisions on their own. That's incredibly useful, and a little terrifying. Without guardrails, a single misconfigured agent can leak sensitive data, run unsafe operations, or go off-script entirely.
This session is a live, hands-on introduction to NVIDIA NemoClaw, an open-source stack purpose-built to solve this problem. NemoClaw adds policy-based security, privacy routing, and behavioral controls to AI agents, giving you real oversight without slowing them down.
I'll deploy NemoClaw twice on stage: once on a Mac Studio to show the local hardware path, and once on NVIDIA Brev to show the cloud path for anyone without a GPU. Every step happens live — from installing the stack to configuring policies to watching an agent hit a guardrail in real time.
Whether you're a platform engineer, a DevOps practitioner, or just someone who wants to run AI agents responsibly, you'll walk away knowing exactly how to get started.