El Bruno

⚠️ This blog post was created with the help of AI tools. Yes, I used a bit of magic from language models to organize my thoughts and automate the boring parts, but the geeky fun and the 🤖 in C# are 100% mine.

Hi!

Have you ever built a chat app where your AI completely forgets what you said five seconds ago? 😅
That’s where persisted threads come in.

In this post, we’ll explore how the new Microsoft Agent Framework for .NET lets you create agents that remember — across multiple interactions, sessions, or even app restarts.

And yes, there’s a demo video too 👉

💬 What Are Persisted Threads?

Every agent created with the Microsoft Agent Framework communicates through an Agent Thread — essentially a container for conversation history and state.

When you persist a thread, you’re saving that entire conversation context somewhere (like a file, a database, or cloud storage) so it can be reloaded later.

Think of it as giving your agent memory — it picks up right where you left off.

💡 Each conversation = one AgentThread
You can save it, restore it, and continue chatting seamlessly.

⚙️ How It Works

The AgentThread object holds the entire conversation context — both user and AI messages.
You can serialize it to JSON, store it, and deserialize it later to continue.

Here’s a simple example 👇

// Create a chat agent
var agent = MyAgentFactory.CreateChatAgent();

// Start a new thread
var thread = await agent.NewThreadAsync();

// Let’s chat!
Console.WriteLine(await agent.RunAsync("Tell me a quick joke!", thread));

// Persist the thread to a file
var serialized = thread.Serialize();
File.WriteAllText("thread.json", serialized);

// Later... reload and continue
var saved = File.ReadAllText("thread.json");
var resumedThread = AgentThread.Deserialize(saved);

// Continue the conversation
Console.WriteLine(await agent.RunAsync("Now tell it as a pirate!", resumedThread));

That’s it — your AI now remembers the previous chat.
No extra work. No complex state management. Just… memory. 💾

🧩 Real Samples to Try

The Microsoft team published official examples in the Generative AI for .NET repository.
You’ll find multiple versions of persisted chat demos:

Each one builds upon the same concept:
store the AgentThread → reload it later → continue the chat.

🏗️ Next Step: Use a ChatMessageStore

When you move beyond local testing, you can use a custom ChatMessageStore to persist data in:

• SQL or Cosmos DB
• Azure Blob Storage
• or even your own API

This is the foundation for stateful AI apps — assistants, copilots, and customer bots that don’t reset with every message.

🧠 Learn More: Generative AI for .NET

Want to go deeper?
Check out the official learning path 👉 aka.ms/genainet

It includes:

• tutorials
• code samples
• agent building blocks
• and guidance for deploying on Azure

This repo is packed with ready-to-run examples that mix .NET + AI beautifully.

💾 TL;DR

—