Foundry for Builders: MCP, Knowledge & Workflow Orchestration

🛠️ Production Is Not Analytics: Understanding Microsoft Fabric's Role

⚙️ Modern production systems are built to process transactions, enforce business rules, and scale reliably.
📊 Analytics systems are built to explore data, aggregate it, and turn it into insight.

Microsoft Fabric exists to cover this second space — with a unified, end-to-end technology stack.

🔧 In this session, we’ll focus on Microsoft Fabric as a platform, with a concise overview of:

• 🗄️ How Fabric stores and organizes data beyond application databases
• 🔄 How ingestion, transformation, and orchestration are handled
• 📈 How analytics, reporting, and exploration are built on top of the same foundation
• 🧩 How Fabric’s workloads are designed to work together as a single system

🧠 Conceptual context will be kept to the minimum necessary to understand why these components exist, with the emphasis placed on what they do and how they are used in practice.

🎯 This session is aimed at developers who want a clear, technology-oriented understanding of Microsoft Fabric and its role next to production applications — without turning into a data platform deep dive.

🛠️ Foundry for Builders: MCP, Knowledge & Workflow Orchestration

Microsoft Foundry introduces a new agent-first way of building AI applications — combining reasoning, knowledge, tools, and workflows into a single, composable platform.

🚀 In this session, we explore Foundry from a builder’s perspective, focusing on how things actually fit together after the Ignite announcements — using the new Foundry UI and APIs.

🔹 What you’ll see

• 🧠 Knowledge-based Agent (RAG)
  Build an agent backed by Foundry Knowledge to answer domain questions using retrieval-augmented generation.

• 🔌 MCP-powered Agent
  Use a Model Context Protocol (MCP) server to securely access an Azure SQL Database — without embedding schemas or credentials in prompts.

• 🔀 Intent-based Workflow Orchestration
  Implement a workflow that detects user intent and routes each request to the correct agent, showing how workflows act as the control plane for multi-agent systems.

🎯 Why this matters

• Clear separation between knowledge, actions, and reasoning

• Practical patterns for multi-agent architectures

• Designed for scalable, governed, production-ready AI solutions

By the end of the session, you’ll understand how to structure agents, MCP tools, and workflows in Microsoft Foundry — and how to apply these patterns to real-world builder scenarios.