Enable agents for enterprises using Agent 365 SDK | OD840
Jeremiah Follis explains how the Agent 365 SDK helps enterprises make custom and third-party AI agents visible, governable, and secure at scale, with a focus on identity, observability, compliance, and risk controls for real business workflows.
Overview
As AI agents move from prototypes into production workflows, the session focuses on the enterprise requirements that show up quickly: being able to discover what agents exist, understand what they do, control what data they can access, and prove compliance through auditing.
Key themes covered in the session description and chapter outline include:
- Enterprise risk areas for agents
- Agent sprawl (agents proliferating without centralized visibility)
- Data oversharing (agents accessing or exposing more data than intended)
- New threat surfaces (agents introducing new security risks)
- Regulatory uncertainty (compliance requirements evolving)
- Treating agents as enterprise actors
- Assigning identity and making actions auditable
- Establishing standards for behavior, data handling, and audit readiness
- Agent 365 as a control plane
- Positioning Agent 365 as a unified place to secure and govern agents
- Emphasis on making agents manageable across an organization
- Agent 365 SDK deep dive
- Incremental integration of observability, governance, and security controls
- Bringing agents under enterprise controls without requiring a full rewrite
- Microsoft security/compliance integrations
- Integration with Microsoft Entra ID for identity
- Integration with Microsoft Purview for compliance and data protection
- Multi-platform considerations
- Agent 365 support across Microsoft platforms as well as AWS and Google Vertex AI