Part 1: Overview and Azure Setup
Start a beginner-friendly, 22-part Microsoft Agent Framework series. See the enterprise IT Service Desk agent you will build, understand its architecture, and set up a keyless Azure foundation with a resource group, Key Vault and a tested Foundry model.
Part 2: A Production-Grade .NET Solution
Set up the DewirideServiceDesk solution the way production teams do — a .slnx solution, a pinned .NET SDK, shared build settings, central package versions, a class library for the agent and an ASP.NET Core API with a health endpoint — step by step in VS Code or Visual Studio.
Part 3: Your First Agent API
Connect the Dewiride Service Desk agent to a Microsoft Foundry model and serve it from ASP.NET Core over AG-UI — Microsoft Agent Framework hosting, keyless sign-in with DefaultAzureCredential, validated settings and streamed answers, tested from an .http file in VS Code or Visual Studio.
Part 4: Multi-Turn Conversations
Make the Dewiride Service Desk agent remember a conversation — an in-memory session store for development, follow-up requests with threadId and parentRunId, and a look at where Microsoft Foundry keeps the history, tested from an .http file in VS Code or Visual Studio.
Part 5: Next.js Chat App
Start the chat UI for a Microsoft Agent Framework agent — create a Next.js 16 app with TypeScript 7 and Biome, pin the newest packages, and give it three settings that are read on every request and checked when the app starts, so one build can talk to any AG-UI agent.
Part 6: Trusted Route Handler
Connect a Next.js chat app to a Microsoft Agent Framework agent the way Microsoft recommends — a trusted route handler that accepts only the user's text, builds the AG-UI request itself, streams the answer back as Server-Sent Events and drops everything the browser shouldn't see.
Part 7: Streaming Chat Page
Build the chat page for a Microsoft Agent Framework agent in Next.js with the AG-UI TypeScript client — answers that stream in word by word, safe Markdown, a Stop button and a New chat button, and one build that any AG-UI agent can reuse by changing environment variables.