Building bots on Microsoft Teams
- Nathan
- Feb 23
- 4 min read
Updated: Feb 27
I recently started down the rabbit hole of running bots and agents in Microsoft Teams. There are many tools, SDKs, and frameworks that can help you get the job done. But which ones should you use? And which ones are outdated or deprecated? That’s what I hope to clear up in this blog post.
Agent Framework, Semantic Kernel, AutoGen
First, you need to build your agent’s code. There are a multitude of frameworks out there which you can choose from. Some popular examples include LangGraph/LangChain and CrewAI, but there are many more. If we want to stay within the Microsoft ecosystem, then there are 3 popular options: Agent Framework, AutoGen, and Semantic Kernel.
AutoGen
AutoGen is "a framework for creating multi-agent AI applications that can act autonomously or work alongside humans." It is known for having simple abstractions for single and multi-agent patterns. It is important to note that AutoGen will not be getting any major new features. In an announcement from October 2025, one of the maintainers stated "AutoGen will still be maintained -- it has a stable API and will continue to receive critical bug fixes and security patches -- but we will not be adding significant new features to it."
Semantic Kernel
Semantic Kernel is "a model-agnostic SDK that empowers developers to build, orchestrate, and deploy AI agents and multi-agent systems. Whether you're building a simple chatbot or a complex multi-agent workflow, Semantic Kernel provides the tools you need with enterprise-grade reliability and flexibility." It is known for having many enterprise-grade features such as state management, telemetry, extensive model support, and more. In this post, Microsoft states that Semantic Kernel is essentially in maintenance mode, and that the team now works on Agent Framework (below).
Agent Framework
The Microsoft Agent Framework is the latest and greatest tool for building agents. Agent Framework is a "comprehensive multi-language framework for building, orchestrating, and deploying AI agents." Agent Framework has been designed to be the next generation of both AutoGen and Semantic Kernel. Agent Framework includes the best features from both AutoGen and Semantic Kernel, while also adding new features on top. Here is an announcement from Oct. 2025 with all of the details regarding Agent Framework, and how you can migrate from either AutoGen or Semantic Kernel.
Teams SDK, M365 Agents SDK, Bot Framework SDK
After you build your agent, the next step is to integrate it with Microsoft Teams. This is another subject where you will find multiple options for getting the job done. Microsoft currently has 2 entirely different and fully supported SDKs for this.
Teams SDK
Let’s talk about the first option: Teams SDK. This SDK started its life known as the Teams AI Library. Then, Microsoft released a successor, which they initially called Teams AI Library v2. But not long after, they rebranded Teams AI Library v2 as simply Teams SDK. In Microsoft’s words, the Teams SDK is “a Teams-centric interface for integrating GPT-based language models and user intent engines. It simplifies the development process by reducing the need to write and maintain complex conversational bot logic ... allows you to focus on business logic rather than learning the intricacies of Microsoft Teams conversational frameworks.”
When to use Teams SDK: In short, if you are building a bot strictly for Teams, then this is probably the best choice as of right now. It is more mature and more feature-rich than the next SDK that we’ll discuss below.
Microsoft 365 Agents SDK
Next, let’s discuss the second option: Microsoft 365 Agents SDK. In Microsoft’s words, “Microsoft 365 Agents SDK provides capabilities for modern agent development, bringing together the creation of conversational agents with conversation management and orchestration. Agents built with the SDK can connect to numerous AI services and clients, including agents created with non-Microsoft software or technology.”
When to use M365 Agents SDK: In short, if you need to support multiple different Bot channels, like Slack, Teams, web, and more, then you will need to choose Microsoft 365 Agents SDK.
Bot Framework SDK
Before we move on, I’d also like to discuss another option you might see out there in the wild, and that is the Microsoft Bot Framework SDK. It is important to note that the Bot Framework SDK is archived by Microsoft and is no longer being updated or maintained. For that reason, I cannot recommend using this SDK. Microsoft’s preferred replacement is the Microsoft 365 Agents SDK (discussed above) which Microsoft calls "the evolution of the Bot Framework SDK."
Microsoft 365 Agents Toolkit
The last thing I wanted to cover in this post was the Microsoft 365 Agents Toolkit. This was previously known as the Teams Toolkit.
This toolkit "makes it simple to get started with app development for Microsoft Teams, Outlook, and Microsoft 365 Copilot using Visual Studio Code." It is available as a free VS Code extension. It has many different features, including, but not limited to:
Create new apps from project templates for common app scenarios.
Save setup time with automated app registration and configuration.
Run and debug to Teams, Outlook, and Copilot directly from Visual Studio Code.
Smart defaults for hosting in Azure using infrastructure-as-code and Bicep.
Test with different configurations such as dev, test, and prod using the environment features.
More to come
There are still more topics to cover with regards to building bots on Microsoft Teams. You need to setup an Azure Bot Service along with a dedicated identity. You must also package up your Bot and add it to your Teams tenant. I plan on writing a part 2 to this post which will cover those topics. So, more to come!