The New Microsoft Copilot

Microsoft Healthcare and Life Sciences Blog:

From Asking Questions to Delegating, Building and Automating

The way we work with AI is evolving quickly.

Microsoft recently introduced the new Microsoft Copilot experience, bringing together new ways to ask, delegate, build and automate work while giving organizations additional controls for managing AI usage and spend.

In our recent webinar, “New Microsoft Copilot App: Which Mode to Use, What It Costs, and Who Gets Access,” we took a closer look at what these changes mean for organizations and the people using Copilot every day.

Rather than trying to recreate the entire session here, this post highlights some of the major topics we covered. For the demonstrations, examples and deeper discussion, watch the accompanying webinar recording.

What we covered

The session focused on four major areas:

The evolution of the Microsoft Copilot experience
The roles of Home, Code and Autopilot
Microsoft Copilot Managed Runtime and enterprise governance
Usage-based AI and the emerging discipline of FinOps for AI

Together, these capabilities point toward a Copilot experience that extends beyond conversational assistance and toward increasingly sophisticated AI-enabled work.

A new Microsoft Copilot experience

Microsoft has announced three major capabilities for the new Copilot experience: Home, Code and Autopilot.

Home brings Chat and Cowork together as a starting point for work. Microsoft is also bringing Word, Excel and PowerPoint capabilities into the Copilot experience, making it possible to create and continue working with Office content without continually rebuilding context across applications.

There is an important distinction between Chat and Cowork.

Chat addresses the conversational interactions many people already associate with Copilot: asking questions, looking up information, creating drafts and staying informed.

Cowork is designed for work you delegate. Rather than simply responding to an individual prompt, Cowork can work through more complex tasks end-to-end and return results for you to review and iterate upon.

Microsoft has also said that the experience is moving toward a point where users won't necessarily need to decide which mode to choose. Users will describe what they want to accomplish and Copilot can route the work to the appropriate capability.

That reflects an important change in how we should think about AI at work. The conversation is expanding from “What can I ask Copilot?” toward “What work can I accomplish with Copilot?”

Code: turning ideas into working solutions

One of the capabilities we spent time discussing and demonstrating during the webinar was Code.

Code is designed to enable people to build solutions by describing what they want to create. Microsoft says the capability is powered by the same underlying technology as GitHub Copilot and is intended to make software creation accessible to both experienced developers and people who may not consider themselves traditional programmers.

During the webinar, we explored this concept through an application-building example based on a requirements document.

The point wasn't that application development suddenly requires no planning or technical discipline. It was to demonstrate how dramatically the starting point can change when AI can help translate business requirements into working software.

Watch the webinar recording to see the example and the resulting application.

Where does AI-generated code actually run?

Making application creation easier immediately raises another question:

How does an organization govern all of those applications?

Microsoft Copilot Managed Runtime is designed to address that challenge.

Microsoft describes Managed Runtime as an enterprise-grade platform for running code within the Microsoft 365 tenant boundary. It provides a common foundation incorporating Microsoft-operated cloud hosting, identity, governed data access and lifecycle management.

Managed Runtime already supports apps created through Copilot Cowork, Copilot Code and Microsoft Copilot Studio.

This creates an important separation between two ideas:

AI makes building easier. Enterprise controls still matter when that software runs.

Microsoft operates the platform, while organizations retain control over the boundaries within which applications operate.

Autopilot: AI that continues working

The webinar also explored Autopilot, which Microsoft describes as a persistent, proactive and personal agent that can continue working even when the user isn't actively interacting with it.

This represents another shift in the AI experience.

Chat is naturally interactive. You ask something and Copilot responds.

Agentic experiences introduce the idea of delegating work that can continue beyond that individual interaction. That opens new possibilities for ongoing research, monitoring, preparation and other work that benefits from persistence.

In the webinar, we discussed what this transition could mean for people who increasingly treat AI not simply as a place to ask questions, but as another resource available to help them get work done.

As AI becomes more capable, cost management matters too

More capable AI workloads can also require different approaches to licensing and consumption.

Microsoft uses Copilot Credits as a common currency for eligible services using usage-based billing. Microsoft describes this consumption model as complementary to fixed subscription licensing, especially as more advanced or multi-step workloads can vary considerably in the amount of work required to complete a task.

That leads directly to another major subject from our webinar: FinOps for AI.

As organizations expand their use of AI, several questions become increasingly important:

What are we consuming?
Which services and scenarios are driving that consumption?
What value are those scenarios creating?
Where should we invest or optimize next?

Control, measure and optimize

A useful way to think about FinOps for AI is through three activities:

Control

Organizations need mechanisms for governing access and spend.

Microsoft's Cost Management capabilities support controls around usage-based AI experiences, including spending policies, limits, alerts and billing methods.

Organizations can therefore begin establishing guardrails around how consumption-based AI services are introduced.

Measure

You can't optimize what you can't see.

Microsoft provides consumption information through its Copilot Cost Management capabilities, giving administrators visibility into Copilot Credit consumption across supported services.

This moves the AI cost conversation beyond simply purchasing technology and toward understanding how it is actually being consumed.

Optimize

Once organizations understand consumption, they can make more deliberate decisions about where AI is producing meaningful value and where different approaches may be appropriate.

That makes FinOps for AI more than a billing conversation. It becomes part of the broader process of deciding where and how an organization wants to scale AI.

A useful way to think about what comes next

Perhaps the biggest takeaway from the webinar is that Microsoft Copilot is expanding beyond a single conversational interaction model.

The emerging experience spans several ways of working:

Ask when you need information or immediate assistance.

Delegate when you want AI to take on a more complex body of work.

Build when you want to turn an idea or requirement into a solution.

Automate when work should continue beyond the immediate conversation.

And surrounding those experiences are the governance and cost-management capabilities organizations need as AI adoption grows.

Watch the full webinar

We've only touched on the highlights here.

In the full webinar, we walk through the new Copilot experience, demonstrate application creation with Code, discuss Autopilot and Managed Runtime, and spend additional time looking at the emerging FinOps capabilities organizations can use to understand and govern AI consumption.

Watch the accompanying video for the complete discussion and demonstrations.

Learn more

Want to explore these announcements and capabilities in more detail?

Check out these official Microsoft resources:

Source: