Generative AI agents run tasks on behalf of human users and often interact with each other across on-premises environments and different cloud providers. This brings new challenges in identity authentication, propagation, delegation, and resource authorization in the overall agentic AI solution. Learn how Amazon Cognito's OAuth2-based identity management, machine-to-machine authentication, combined with Amazon Verified Permissions fine-grained authorization can enable secure delegation patterns for AI agents, while preserving human identity and consent, agent machine identity, and other request context throughout the agent chain. We will walk through real-world examples with agents built on Amazon Bedrock or other frameworks.
2 sessions
- Code talk400 - ExpertIdentity and Access ManagementGenerative AIAgentic AI4:00 p.m. Monday, Jun 16Monday, Jun 16Session Type: Code talkLevel: 400 - ExpertTrack: Identity and Access Management
- Code talk400 - ExpertIdentity and Access ManagementGenerative AIAgentic AITuesday, Jun 174:00 p.m. Tuesday, Jun 17Generative AI agents run tasks on behalf of human users and often interact with each other across on-premises environments and different cloud providers. This brings new challenges in identity authentication, propagation, delegation, and resource authorization in the overall agentic AI solution. Learn how Amazon Cognito's OAuth2-based identity management, machine-to-machine authentication, combined with Amazon Verified Permissions fine-grained authorization can enable secure delegation patterns for AI agents, while preserving human identity and consent, agent machine identity, and other request context throughout the agent chain. We will walk through real-world examples with agents built on Amazon Bedrock or other frameworks.Session Type: Code talkLevel: 400 - ExpertTrack: Identity and Access Management