Friday 12:30 PM - 1:40 PM · Room A1

Copilot Studio vs. Azure AI Foundry

Peter Carson

President

AI powered agents feel like they're coming everywhere, but how do you light up your own data in them not just for staff, but also for external guests? How can you incorporate an AI Agent into both scenarios securely?

We'll start the session with a level set on how AI really works, and in particular Retrieval Augmented Generation. A combination of Search and Large Language Models, it lets your chat interface answer questions it hasn't been trained by leveraging search. Built properly it also respects security, and only answers from content the user has access to, which can live in Dataverse, Microsoft 365, or any other search accessible repository.

One approach we'll cover is Copilot Studio, a low-code environment where you can build intelligent chat experiences using ready-made large language models, a dialog manager, and numerous data connectors. These can be published and licensed in a variety of ways, and can enforce security against your enterprise data.

The other approach is more code focused in Azure AI Foundry. If you want you can get into details around search population, chunking, AI models, and UI.

We'll also cover consumption based pricing from Microsoft for both.

Feeling faint of heart with all this tech talk? Don't despair. Yes we'll be showing code, but we'll also go over real-world business cases, and the solution architectures that leverage the Microsoft AI stacks.