Architecting & Securing AI Agents

Architecting & Securing AI Agents

Join us for a meetup where we dive into the world of AI agents! Come to Check Point with AI21 Labs and explore how to architect, implement, and secure AI agents through live coding and insightful discussions.

Nitzan Cohen, AI21 labs: Ever feel like there's a new agent dropping every time you refresh Twitter? You're not wrong!
This talk takes you on a breezy tour through the latest trends in the rapidly evolving landscape of AI agents. We'll begin with the popular ReAct (Reasoning and Acting) framework, and understand where it shines and where it stumbles. And we'll continue with advanced planning mechanisms that help agents think ahead, memory management, and optimization tricks that make them actually run in a reasonable amount of time.

Benny Zemmour, Check Point, will explore the security vulnerabilities of AI agents and demonstrate practical defense strategies. You'll see a live demonstration of how seemingly innocent inputs can be weaponized to manipulate AI systems into performing unauthorized actions. We'll then build a comprehensive security framework—from supply chain considerations to runtime protection—that developers can implement to create resilient AI agents. These techniques apply across various AI agent deployments and will help you stay ahead of this rapidly evolving threat landscape.

Host or Publisher Check Point