Craig Stanley
Home / Capabilities / Home, Code and Autopilot / The philosophy: Copilot as the OS for work

The philosophy: Copilot as the OS for work

My reading of Microsoft's September 2026 Copilot announcement: one place to ask, delegate, build and run agents, and what that means for decisions.

· 3 min read · Craig Stanley
In short, explained

Microsoft wants its helper to be the first thing you open at work. From there you can ask questions, hand over jobs, make small tools and have a helper that keeps working when you're away.

In September 2026 Microsoft reorganised Copilot around Home, Code and Autopilot. My reading is that Microsoft wants Copilot to be the place work starts from, a bit like an operating system. Most of it is still in early access or preview, so it's a direction for now.

The 25 September 2026 announcement positions the Copilot app as the entry point for asking (Chat), delegating (Cowork), building (Code, with Copilot Managed Runtime) and persistent agents (Autopilot), grounded in Microsoft IQ and paid partly through usage-based billing. "OS for work" is my framing; Microsoft doesn't use the phrase. Availability is staged: Frontier, private preview and preview.

What Microsoft announced

On 25 September 2026 Microsoft published "Introducing the new Copilot with Home, Code and Autopilot" on its official blog. In its own words, the Copilot app "now has three new capabilities".

Home is the starting point, "where Chat and Cowork come together", with Word, Excel and PowerPoint built into the experience through what Microsoft calls Office in Copilot. Chat is for quick questions. Cowork is for work you delegate: you define a task and it runs to a finished result.

Code lets people describe an app, tracker, dashboard or automation in plain language and have Copilot build it. Microsoft says it's powered by the same technology as GitHub Copilot and runs in a sandbox that can be hosted in your tenant, on a new hosting layer called Copilot Managed Runtime.

Autopilot, which was previously called Scout, is described as "a persistent, proactive and personal agent". You give it a name, a role and a goal. It lives in your tenant with its own identity and can be @mentioned in Teams and Outlook like a colleague.

Where each piece stands

This is the part I'd keep in front of anyone planning around the announcement. As of the blog post and Microsoft's related documentation:

CapabilityStatus in the announcement
Home and CodeRolling out first through Microsoft's Frontier programme; Microsoft says Code reaches broad availability "in the coming weeks" and preview for Microsoft 365 Premium and Pro subscribers later in 2026
AutopilotExpanding to private preview at the end of September
Copilot Managed RuntimePreview
Today (in Home)Private preview in October
@Copilot in TeamsPrivate preview by the end of September
CoworkGenerally available, according to Microsoft Learn

So most of what was announced is early access or preview. I'd treat it as a strong signal of direction and plan for it, and I wouldn't yet build a business process that depends on it.

Why I call it an operating system

"OS for work" is my phrase, and Microsoft doesn't use it. I use it because the announcement has the shape of an operating system. There's a home screen. There's a way to run tasks in the background (Cowork and Autopilot). There's a way to build and host small programs (Code and the Managed Runtime). There's a shared layer of context underneath, which Microsoft calls Microsoft IQ, and an app store of sorts in the new plugin registry. And there's a billing model to match: Microsoft says everyday AI stays on the per-user licence, while Cowork, Code and Autopilot run on usage-based billing.

What it means for decisions

If Copilot becomes the place work starts, more small decisions will be made inside it, often by agents acting on someone's behalf. Autopilot is the clearest case. Microsoft's example is an agent that runs a supplier review: building the schedule, handling preparation and follow-ups, and chasing stakeholders. Every one of those steps involves small choices.

That's the reason this site focuses on decisions. When an agent keeps working while you sleep, I want to know which decisions it's allowed to make, with what limits, and how anyone would find out if it got one wrong. The Risk section is where I work through that.

A worked example

This is an illustrative scenario. A finance team's weekly supplier review takes about three hours of a coordinator's time. An Autopilot agent could, on Microsoft's description, build the schedule, collect updates and draft the pack. The decisions that remain with people are which suppliers to escalate and why. I'd write those down as explicit rules before giving the agent the rest of the job.

What I'm still checking

I haven't used Autopilot or Code myself, because both are in early access. Everything above about them comes from Microsoft's own descriptions. I'll update this page when I've tried them.

Sources

Read next

A question to take awayWhich of these do you already pay for and not use?

About me

Craig Stanley

Microsoft AI consultant and technical architect, based in Whitley Bay. Over the last few years I've delivered Microsoft 365 Copilot, Copilot Studio agents, Microsoft Foundry (formerly Azure AI Foundry) work and governance for UK public sector and financial services organisations.

What interests me is the decision underneath the tool: what it costs, what it risks, and whether a small, transparent model can make it better. I write the methods up here and on Substack so anyone can use them.

I write this site to learn in public: explaining each idea simply is how I check I understand it. Why I write this site.

Find me