An always-on AI agent is coming to your Microsoft 365. Decide the rules first.

If your business runs on Microsoft 365, here is something worth watching. Microsoft is putting always-on AI agents inside the tools you already pay for, and they don't just answer questions. They do things, across your files and apps, on their own. The capability is genuinely useful. It also arrives whether you plan for it or not, which is exactly why it is worth thinking about now rather than the day it switches on.

What's actually arriving

At its Build 2026 conference in June, Microsoft introduced Scout, which it describes as an always-on agent. Satya Nadella framed the whole category, which Microsoft calls "Autopilots," as "always-on agents that work autonomously, with their own identity, and act on your behalf." Scout works across Teams, Outlook, OneDrive and SharePoint, and, notably, it joins group chats and manages email threads as a participant rather than sitting quietly in a sidebar.

You may never switch on Scout specifically. That is not really the point. The point is the direction: the everyday tools a huge number of small businesses already run on are about to be able to carry out work, not just draft it.

Why this is a real step up

Most small businesses already pay for Microsoft 365 and use a fraction of what they pay for. An assistant that can pull from your actual documents and carry out a multi-step task, rather than just chat about it, is a meaningful jump in usefulness. For a small team without spare hands, "it went and did the thing" is worth a lot more than "here's a draft for you to do the thing."

So the opportunity is real, and it is sitting inside a subscription you already have. That is the good news, and it is genuinely good.

"Can act across your files" is a sentence to read twice

Here is the same sentence from the other side. The very capability that makes it useful, that it can reach across your files and apps and act, is the one that deserves a moment's thought before you turn it loose. An agent that can touch everything, on its own, from day one, is not what you want.

This is where a Microsoft 365 agent meets a simple governance idea: capability and access are two different decisions. How clever the tool is, and what it is allowed to reach and do on its own, are not the same question, and you get to answer the second one yourself. The same agent is low risk or high risk depending entirely on what you let it touch unchecked.

Treat the rollout as a governance decision, not just a new feature

The good news is you don't have to choose between ignoring it and switching it on blindly. Microsoft is building the controls for exactly this. Alongside Scout it introduced Agent 365, which treats these agents as managed identities with their own sign-in, so administrators can review, monitor and block agents from the Microsoft 365 admin centre. Microsoft has also said the full tenant-level controls are still rolling out through 2026, which is all the more reason to decide your own rules deliberately rather than accept whatever the default turns out to be.

None of that requires you to become an expert. It requires a decision: what is this agent allowed to work on first, and what is it explicitly not allowed to touch. That "adopt it, but on purpose" is exactly the sort of thing we help owners think through at Handiwork, and it is the same discipline we bring to any workflow we build and put live.

Start it in one corner

So when an always-on agent lands for you, don't flip it on across everything at once. Decide the one area it is allowed to work in first, something low-stakes, and what stays off limits. Let it prove itself there, and widen from there as it earns your confidence. The tools you already pay for are about to get much more capable. Meeting that on your own terms is the whole game.

Ready to find out where you stand?

If you would like help adopting this safely when it lands, the free AI Readiness Check is a good first step. It takes about five minutes and there is no pitch at the end of it.