As you look to enable builders to consume Generative AI models, most teams will want the capability to utilize different models across various different use cases. This can become incredibly difficult to govern, as each platform has its own unique configurations, governance controls, and methods of access/permissioning.
An AI/LLM Gateway is a specialized type of layer 7 proxy that is specifically built with governing LLM, MCP, and agentic workload access in mind. There are numerous operational benefits, but the primary security benefit is that we can have a consistent security control interface across various consumers and models.
This session will explore the usage of AgentGateway, a Linux Foundation project that provides an open source AI Gateway. AgentGateway also cleanly integrates with Kubernetes, allowing you to fully manage its configurations declaratively in code as Kubernetes Custom Resource Definitions (CRDs). We will then apply features of AgentGateway to solve security challenges around LLM access in live demos! Attendees can expect to leave with a greater understanding of LLM Gateways, AgentGateway, and concrete use cases.
Who should attend?:
Application Security, DevSecOps, Security Engineers, or anyone interested in learning about the security controls/benefits of an LLM/AI Gateway
Learning objectives:
- Learn about challenges in governing LLM usage across consumers and models
- Understand the purpose and usecases of an LLM Gateway
- Learn how to secure access from CLIs, Agents, and Applications to LLMs and MCPs with real implementation examples
This session supports concepts from the SANS SEC540: Cloud Native Security and DevSecOps Automation. To learn more about this course, browse upcoming runs, and access your free preview, CLICK HERE!
