
Microsoft Copilot: What Home, Code and Autopilot Add
Microsoft Copilot adds three ways to work: Home, Code and Autopilot. Here is what each does, how app hosting works and when previews start arriving.
Microsoft Copilot is expanding from a place to ask questions into a workspace for documents, small applications and ongoing assignments. Microsoft's September 25 announcement introduces Home, Code and Autopilot, with access arriving in stages. The interesting question is how these different kinds of work fit together.
- Home brings Chat, Cowork and editable Office files together.
- Code builds applications and workflows from natural-language requests.
- Autopilot is a persistent agent; its private preview expands at month-end.
- A separate Managed Runtime public preview supplies enterprise app hosting.
What changes in Microsoft Copilot Home and Code?
According to Microsoft, Home connects conversation with work in Word, Excel and PowerPoint. Code adds tools for creating things such as trackers and internal dashboards, using the underlying technology behind GitHub Copilot. Home and Code begin arriving through the Frontier program in the coming weeks; this announcement does not put every feature into every account immediately.
Microsoft describes Autopilot, previously called Scout, as an agent that can maintain ongoing work. Our reading is that the practical distinction is persistence: a recurring assignment needs continuity between sessions, whereas a single document request has a clearer stopping point.
Where do the generated apps run?
Microsoft's separate Managed Runtime announcement describes hosting within the organization's Microsoft 365 tenant boundary. Entra identity controls access, policies govern connections to data, and Git supports version and source control. Administrators receive a central inventory for reviewing applications, access and usage.
Its developer tools also cover building, previewing and deploying applications. That gives professional developers a route into the same hosting environment, rather than making the platform depend exclusively on conversational app creation.
For teams following enterprise AI developments, this operational layer deserves attention. Creating a useful prototype is one milestone; knowing who owns it, who can open it and how to recover an earlier version determines whether colleagues can rely on it.
What should teams evaluate first?
A focused pilot could track a simple internal request from creation through revision and retirement. Our suggested test is ordinary but revealing: can another teammate understand the app and maintain it after its original builder moves on?
That complements the engineering questions in our coverage of GitHub's agent-assisted Rust migration. For Copilot's broader audience, the opportunity is making useful software easier to create and easier to operate responsibly.
Sources: Microsoft Copilot announcement — September 25, 2026; Microsoft Copilot Managed Runtime overview — September 25, 2026. These are vendor descriptions of preview capabilities, not independent performance evaluations.
More AI Stories

Strands Decider 2B: What Amazon's Free Decision Model Does
Strands Decider 2B is a free 2B-parameter decision model that picks options in about 115ms on a single GPU. Here is how it works and where it fits.

Meta Muse Gadgets SDK: Build Your Own AI Agent Hardware
Meta's open-source Muse Gadgets SDKs bring its Muse agent to Raspberry Pi and ESP32 builds, and 5,000 free Home Link dongles are going to subscribers.

AstaBrief 8B: Ai2's Open Model for Cited Science Reports
AstaBrief 8B is Ai2's Apache 2.0 open model that writes cited research reports in 51 seconds, 3.5x faster than its Claude pipeline. Here is how it works.
