Real World MCP: Integrating MCP within the Azure Ecosystem

9:00 午後 - 10:00 午後 (UTC)

In this session, we explore how to harness the power of the Microsoft Docs MCP through a custom-built Web Client. The project features a Next.js web application that functions as an intelligent chatbot for the Microsoft Learn platform. By connecting to the official Microsoft Learn MCP Server, the app retrieves authoritative documentation and uses the Azure OpenAI API to synthesize clear, concise answers—turning static content into dynamic, conversational learning. This is more than just a chatbot—it's a personalized gateway to Microsoft Learn, built for developers who want fast, accurate insights with a modern UX.

9:00 午後 - 10:00 午後 (UTC)

This session unveils a powerful integration between Azure AI Foundry Agents and a customized MCP Server—designed to transform natural language requests into fully generated Azure Architecture Diagrams and Design Documents. Hosted as an Awesome AZD Template, the project exemplifies how declarative infrastructure meets intelligent design. We’ll walk through the enhanced version of the solution, showcasing its capabilities and the exciting potential for future expansion. Whether you're architecting cloud-native systems or streamlining design workflows, this is a glimpse into how AI can accelerate and elevate technical delivery—one prompt at a time.

9:00 午後 - 10:00 午後 (UTC)

Streamlit Meets Scenarios: Building Adaptive AI for the Enterprise Enterprise needs are rarely static—they evolve fast and unpredictably. In this session, we demonstrate how to build and deploy a Streamlit-powered AI application tailored for dynamic use cases, using scenario-based templates like Coding Assistant, Document Summarizer, and more. The solution serves both Web Clients via an MCP Wrapper and MCP-native clients like Claude.ai, offering seamless integration across platforms. The user interface is intuitive and highly configurable—users can select models, modify system prompts, download JSON logs, and manage multiple sessions effortlessly. Session orchestration is powered by FastAPI, exposing a rich set of tools to start, monitor, and control interactions—whether initiated from the web or directly through the MCP Client. Under the hood, Azure Container Apps and DAPR deliver scalable hosting and resilient performance, making this architecture ideal for demanding enterprise environments.

9:00 午後 - 10:00 午後 (UTC)

Developing a Windows AI App powered by Azure-hosted MCP The future of AI demands more than innovation—it requires thoughtful governance, measurable metrics, responsible usage, and yes, even rate limits. That’s not a punchline—it’s a reality. Organizations today have a spectrum of options to architect and manage their own AI ecosystems. In this session, we showcase a sleek Windows-based AI application that empowers authenticated users with seamless access to tools and models hosted on an Azure Web Apps–powered MCP server. Built on .NET, this solution strikes the perfect balance between open AI connectivity and controlled access to content. It leverages Microsoft Azure’s robust security and performance stack, while integrating Entra ID for device and app management—enhanced further with Intune. This is integration done right: native, secure, and scalable. Welcome to the new standard of AI-powered interoperability.