Best Practices for Always-On AI Agents Using Copilot Studio and Power Platform
Presented by Shivani Pradhan at Microsoft Ignite 2025, this session delivers practical insights on architecting always-on AI agents with Copilot Studio and Power Platform to ensure business continuity and scalability for critical workloads.
Best Practices for Always-On AI Agents Using Copilot Studio and Power Platform
Speaker: Shivani Pradhan
Event: Microsoft Ignite 2025
Overview
This session focuses on best practices to architect and deploy always-on AI agents that support mission-critical workloads, leveraging Microsoft services like Copilot Studio, Power Platform, and Dataverse.
Key Topics Covered
- Power Platform and Azure Outage: Lessons on platform resilience and the business risks associated with major cloud outages, with a real-world case from the Norway Azure event.
- Continuous Operations: Strategies to ensure uninterrupted workflow execution in frontline and executive scenarios.
- Copilot Studio in Insurance: Exploring how Copilot Studio enables resilient agent environments, including a look at agent architecture specific to the insurance industry.
- Failover and Data Loss: Explanation of typical failover processes, caution about the one-minute data loss during transitions, and emphasizing responsibility before triggering failover.
- Resilient Deployment: Introduction to deployment methodologies that improve safety and rapid recovery.
- Monitoring and Observability: Best practices for real-time monitoring, scaling, and automation to ensure agents are robust against failure.
- API Automation and SSDR: Automating secondary site disaster recovery (SSDR) through APIs while stressing the importance of human oversight for critical processes.
- Compliance Drills: Using simulated production failovers to train teams and validate preparedness for real disruptions.
Useful Resources
Recap Table
| Topic | Key Takeaway |
|---|---|
| Platform Resilience | Importance of robust architecture during outages |
| Automation | Use APIs for SSDR but include human checks |
| Observability | Monitor health and automate scaling |
| Compliance & Testing | Run drills to maintain readiness |
Conclusion
Through real-world examples and actionable best practices, this session helps technical makers and architects ensure their AI solutions on Microsoft platforms can withstand critical failures and maintain continuous business operations.