Speakers
Overview
Most enterprises have approached AI security through the lens of Assistants—tools that summarize information or generate content. In that model, the main concern is data access and privacy, ensuring that no enterprise data leaks out.
With the rise of AI agents that can take autonomous action, the risk profile expands to include not just the information agents can see but also the actions that they can take in enterprise applications.
In this talk, Glean’s Senior Solutions Engineer Hamza Sharif walks through a framework for securing the entire agent lifecycle- from data access to permissions to observability to AI security and more. By grounding the conversation in software development best practices, Hamza offers a clear and structured way to reason about agent security.
Watch now to learn:
- How to mitigate risks like jailbreak attempts, prompt injection attacks, and unauthorized actions
- How software design principles of zero trust and shift left apply to agents
- How to build agent guardrails, ensuring they operate within the bounds of the original goal
- Why data readiness and governance are foundational to safe AI deployment
