The Microsoft Teams platform mission is to build the best collaborative platform in the world. We want to make it easy for developers to build agents that participate seamlessly in chats, channels, and meetings, helping teams turn ideas into action in real time. By bringing agents directly into the places where collaboration happens, we can help ensure work is completed in the moment rather than getting lost in a stream of conversation. Everything we build is designed to help developers create collaborative experiences that empower people and agents to work together more effectively.Today we're launching Building Agents for Teams, a new monthly blog series here on the Microsoft 365 developer blog dedicated to sharing what we're building, why we're building it, and how developers can build on it. Expect a mix of deep technical guidance, product updates, customer and partner stories, and perspectives on the future of collaboration.If you build agents, or you want to, this blog series is for you.The agent opportunity in TeamsMost agents today are confined to standalone apps and tabs, forcing work into a series of disconnected 1:1 interactions. People must seek out the right agent, enter its environment, and adapt to its way of working instead of having agents participate naturally in existing workflows. That friction may be tolerable for intensive tasks like research or coding, but it quickly adds up across the countless small moments of work that happen every day—checking status, finding data, granting approvals, and bringing the right people together.In Teams, an agent behaves like another person you collaborate with. You @mention it, add it to a channel, give it a task in plain language so it answers in the thread, and you bring it into a meeting when you need it. There’s no manual or training cost, because people already know how to work with each other, and your agent works the same way.That means agent adoption happens the moment your users log in for the day. The hard part of distribution - getting your agent in front of real people in the flow of real work - is already solved.That’s the opportunity for developers: broad user reach and zero learning curve. You can build something useful and put it where work is already happening.Why should you build an agent now?A year ago, you could demo an agent, but today you can ship one that completes real tasks. LLMs have made major leaps forward, the latency isn’t a blocker, and the cost keeps falling.Model Context Protocol gives agents a common way to reach tools, skills and markdown format memories and functionality, and agent-to-agent communication gives them a way to work together. These aren’t Microsoft-centric ideas, they’re standards for how the whole industry connects agents, and the Teams platform speaks them natively.We are at a unique junction where the technology, standards, and demand are aligned at the same time, making this the perfect time to build collaborative agents.The value of collaborative agentsTeams is the right home for agents because agents belong where people already collaborate, inside the threads and meetings where the team lives, rather than in a sidebar you have to remember to open. To enable seamless collaboration, we support two kinds of interactions: Private interactions are one-to-one, a person and an agent solving a problem together. This is the familiar shape, and we make it excellent: fast, contextual, and private to that person. You can privately interact with an agent in Teams 1:1 chats or with targeted messages grounded in the confines of a group chat, channel, or meeting. Public interactions are where it gets interesting. Picture an agent in a group chat, in a channel where the whole team can see it work, or in a meeting where it contributes while everyone watches. Here the agent isn’t a private tool, it’s a participant. Designing for both is harder than designing for one, but the potential value increases exponentially. The agents that matter most will be the ones your whole team can work with, the way you work with each other. We’re building the platform so you can make those agents without reinventing the hard parts yourself.Build an agent with the Teams SDKThe Teams SDK is your one-stop-shop for building your collaborative agent. Authentication, event routing, and the Teams-specific details are done for you. You just write the logic that makes your agent worth using. You can build across TypeScript, C#, and Python. It supports Model Context Protocol for shared tools and memory, and agent-to-agent communication for multi-agent work. You get native support for Adaptive Cards, message extensions, embedded web apps, dialogs, and Microsoft Graph. Pick your language, scaffold a project, and ship something small that does one job well, then put it in front of your team. The fastest way to understand this platform is to build on it, and the fastest way to build is to start now:npm install -g @microsoft/teams.cliteams project new typescript my-agentWhat's nextThis is a series, so here’s some of what’s coming. We’ll go deep on building for public interactions, the channel and meeting scenarios that make Teams different. We’ll cover Model Context Protocol and agent-to-agent patterns with real code. We’ll walk through publishing and reaching users at scale. We’ll break down privacy and compliance challenges. And we’ll share what we’re learning from the developers already building with us. We’re building a collaborative agentic platform, and we’d rather build it with you than for you. We encourage you to give us your feedback and be part of the future we want to work in. Welcome to Building Agents for Teams!Resources Agents in Teams Teams SDK welcome and overview Code samples