Microsoft Events presenters Ken Archer, Sarah Bird, and Florent Ricci deliver insights on monitoring, optimizing, and securing enterprise-scale AI agents using the Microsoft Foundry Control Plane at Ignite 2025.

AI Operations: Real-Time Monitoring and Governance with Microsoft Foundry Control Plane

Speakers: Ken Archer, Sarah Bird, Florent Ricci
Session: Microsoft Ignite 2025 (BRK205)

Overview

This session dives into observability, governance, and security for organizations deploying AI agents at scale, using Microsoft Foundry's Control Plane. Experts from Microsoft break down tools and strategies for maximizing reliability, compliance, and operational efficiency when running advanced AI workloads in production.

Key Topics

Two-Dimensional Security Approach for AI Agents

• Explains Microsoft's principles of securing both the external and internal access vectors for AI agents.

Foundry Control Plane Capabilities

• Walkthrough of real-time visibility and control mechanisms for managing AI agents.
• Provides actionable dashboards and observability features for tracking agent performance and lifecycle.

Agent Risk Management

• Highlights how to address risks like external data exposure and unintended internal access.
• Demonstrates creating guardrails and adding controls to mitigate risks such as prompt injection and handling of personally identifiable information (PII).

Observability and Human Oversight

• Introduces the observability model, which brings transparency to agent behaviors while empowering administrators with human-in-the-loop oversight for trustworthy decision-making.

Continuous Evaluation on Real Production Data

• Describes how continuous evaluation helps identify issues and ensures ongoing agent robustness in dynamic environments.

Integrated Security with Microsoft Defender

• Details deep threat detection integration, enabling automated security responses and comprehensive auditing.
• Reviews how policy alerts and Defender work together within the Foundry ecosystem for end-to-end protection.

Compliance and Governance

• Shares a practical playbook for enforcing compliance standards and governance best practices when running mission-critical AI.

Case Study: Stanley Black and Decker

• Presents a real-world example of deploying, monitoring, and governing AI agents in production using Microsoft Foundry.

Additional Resources

• Manage GenAI Lifecycles at Microsoft Ignite
• Microsoft Ignite Homepage

Conclusion

Whether you are scaling AI workloads or looking to harden your current deployments, this session offers actionable guidance for deploying robust, compliant, and secure AI agents using Microsoft Foundry's Control Plane.