AI-powered agents are becoming ubiquitous, but how can you securely leverage your own data—whether for internal staff or external guests? This session explores how to safely light up your data using tools like Power Automate, which allows user profile injection to inform the AI who is asking, and Power Pages, which support both anonymous and authenticated websites. We’ll show how to integrate AI agents into both scenarios—securely and effectively.

We’ll begin with a level-set on how AI really works, focusing on Retrieval Augmented Generation (RAG)—a powerful combination of search and large language models. RAG enables chat interfaces to answer questions beyond their training by leveraging internal content repositories. When implemented correctly, this approach respects security and only returns answers based on data the user is authorized to access—from Dataverse, Microsoft 365, or other integrated sources.

Next, we’ll explore Copilot Studio, a low-code platform for building secure, intelligent chat experiences using pre-built language models, a dialog manager, and robust data connectors. These experiences can be deployed directly to Power Pages while maintaining enterprise-grade access controls.

We’ll then transition to Azure AI Foundry, examining how it complements or replaces Copilot Studio solutions with a component-based approach. AI Foundry lets you assemble full AI solutions using Microsoft and third-party tools—extending your capabilities across the Azure AI ecosystem.

Finally, we’ll demystify Microsoft’s often confusing licensing and pricing, helping you make informed choices aligned with your organization’s needs.