Microsoft has pushed a major update to Scout, its AI assistant woven into Windows and Microsoft 365. The headliner is live model switching—you can now hop between AI models mid-conversation without losing context. Alongside that, Scout gains tighter links to Teams and Microsoft To Do, a larger context window, adjustable reasoning controls, and expanded tools for developers through the Model Context Protocol (MCP).

Live Model Switching Comes to Scout

Until now, Scout tethered you to a single AI model for an entire session. If you started with a fast, lightweight model for quick answers and later needed deeper reasoning, you had to start over. The latest update erases that friction. A model selector now lives inside the chat interface, letting you swap on the fly. According to Microsoft’s release notes, the switcher surfaces all available models—including GPT-4, GPT-4 Turbo, and more efficient options—and preserves the conversation history across switches.

In practice, you might ask Scout to draft an email with a speedy model, then flip to a more capable one to analyze a spreadsheet attached to the same thread. The assistant re-processes the task with the new model’s strengths without dropping prior context.

Adjustable Reasoning and Bigger Context

The update also introduces sliders to control how much “thinking” Scout does before responding. Microsoft calls these reasoning controls. Crank it up for math, coding, or multi-step planning; dial it down when you just need a snappy summary. This mirrors similar controls in other AI tools but is now a first-party lever inside Scout.

Context windows—the amount of conversation history and documents the AI can hold in mind—have been expanded significantly. Microsoft didn’t put a hard token number in its announcement, but early testers report handling meeting transcripts over 30 minutes long without Scout forgetting earlier parts. That makes it viable for deep-dive research sessions or long troubleshooting chats.

Deeper Teams and To Do Integration

Scout can now act directly inside Microsoft Teams channels. You can ask it to summarize a channel’s activity from the last week, draft a message based on recent posts, or even trigger a follow-up meeting. On the task management side, Scout can create and manage Microsoft To Do items from natural language prompts—turn a chat message into a task with a due date and a reminder, no copy-paste required.

These integrations rely on improved multi-step automation. For example, you can say, “Summarize the decisions in the ‘product launch’ channel and create a task for each action item assigned to me.” Scout will pull the channel summary, identify action items, check which ones mention you, and generate corresponding To Do tasks.

What This Means for Your Workflow

For everyday users

The model switching makes Scout feel more like a toolbox than a single tool. You’re no longer forced to decide upfront whether you want speed or depth. The Teams and To Do hooks cut down on app-switching fatigue—fewer clicks between reading a message and capturing a reminder.

For power users

Adjustable reasoning is a quiet game-changer. If you’re coding, you can push Scout to think through edge cases; if you’re brainstorming, you can keep it breezy. The larger context window means you can feed it entire documents or long email threads and get coherent analysis back.

For admins and developers

Microsoft has added MCP (Model Context Protocol) tools, allowing third-party services and internal line-of-business apps to connect directly to Scout. Admins can govern which models users can switch to via policy, and developers can build custom actions that Scout orchestrates. For instance, an IT team could wire Scout into their ticketing system to auto-generate incident reports from a Teams chat.

How We Got Here

Scout debuted earlier this year as Microsoft’s AI assistant that’s more deeply integrated into the OS than the web-based Copilot. Early feedback praised its local app control—it can adjust Windows settings, launch apps, and manage files—but users wanted more flexibility and enterprise-grade hooks. This update lands just as competitors like Google Gemini and ChatGPT are also rolling out agentic features, making model flexibility and app integration table stakes.

The multi-step automation in Scout builds on Microsoft’s Power Automate lineage, while the MCP support signals a broader push to make Scout a hub for AI-driven workflows rather than a standalone chatbot.

How to Get Started with the New Features

The update rolls out automatically to Scout users on Windows 11 via a background service update. If you haven’t seen it, check manually: open Scout’s settings, choose “About,” and click “Check for updates.” The model switcher appears as a drop-down next to the chat input field; reasoning sliders live in the advanced settings panel.

To use Teams integration, you must be signed into a Microsoft 365 work or school account with Teams access. The To Do connection works with both personal and work accounts. MCP tools require an admin to enable them in the Microsoft 365 admin center under “Scout” settings.

Microsoft warns that live model switching is still marked as “preview.” Some models might respond differently to the same prompt, so review critical outputs before sharing.

What’s Next for Scout

Microsoft’s roadmap points to more agent-style automations—Scout will eventually execute multi-step plans across apps without you watching. Rumors of deeper Excel and Outlook integration swirl, and the MCP gateway hints at a forthcoming plugin marketplace. For now, the live model switch alone makes Scout a much nimbler assistant, and the Teams tie-in plants it firmly in the collaboration stack. Keep an eye on Microsoft Build for likely expansions.