Deep Dive

Revised with .NET examples — A newer version of this article, covering both Python and .NET, is available as part of the MAF v1: Python and .NET series: MAF v1 — 09-workflow-executors-and-edges, MAF v1 — 10-workflow-events-and-builder.

Stream tokens as they arrive and reuse a session across turns so the LLM sees the full conversation. ~60 lines of code per language.

Serialize an AgentSession to JSON, persist it, reload in a fresh process, and have the agent pick up exactly where it left off.

Build the smallest useful Microsoft Agent Framework program: one LLM call, one response, and the same shape in Python and .NET.

Install uv, .NET 9, Docker, OpenAI keys, and a verify script so Chapter 1 starts with both Python and .NET environments ready.

Give an agent a callable function, let the LLM decide when to use it, and compare Python and .NET tool loops side by side.

Every vendor selling software right now claims their product is “agentic.” I’ve seen chatbots with a system prompt called an agent. I’ve seen a scheduled Python script described as autonomous AI. I’ve also shipped actual agents to production — at an insurance company, handling FNOL triage, policy lookup, and claims routing. The gap between what gets marketed as an agent and what you’d actually build is significant.

If you write JavaScript or TypeScript professionally, you’ve used npm. It ships with Node.js, it’s everywhere, and for years it was the only game in town. But after running npm on real-world projects — enterprise monorepos, Next.js apps, Angular dashboards, CI pipelines that bill by the minute — I can tell you: npm is holding you back.

macOS has had built-in dictation since Monterey. It is fine — press and hold a key, speak, done. But it requires Apple’s servers (unless you download the enhanced on-device model), only works in some apps, and you have zero control over punctuation, formatting, or hotkeys.

If you have ever tried to call the Microsoft Graph API, integrate with SharePoint, or add “Sign in with Microsoft” to an application, you hit the same wall: you need an App Registration. It is the first thing you do, and if you get it wrong, nothing else works.

Every developer hits the same wall eventually. You are building a webhook integration, testing an OAuth flow that demands a public redirect URI, or trying to show a client a work-in-progress without deploying it. Your app is running fine on localhost:5000, but the outside world cannot reach it.

The infrastructure overhead in microservices projects is real. Setting up databases, wiring service discovery, getting distributed tracing working, managing connection strings across environments—this work hits before you’ve written a line of business logic. On more than one project, I’ve watched the first sprint disappear into Docker Compose files and appsettings drift.