Microsoft Insights: Issue 8

Table of contents

Copilot can now speak up in Teams meetings

PowerPoint Copilot is learning your brand

You can now tell AI to leave the meeting

Teams’ meetings are getting a reshuffle

Microsoft is putting more guardrails around AI

Data protection is stretching beyond Microsoft 365

On your radar: Power Apps pricing

That's September

Welcome to Issue 8 of Microsoft Insights

This month, there's a noticeable shift in Microsoft's approach to AI. Copilot isn't just sitting there waiting for someone to type a prompt. It's starting to recognise when it can help, work more independently and use more of the context around you.

In Teams, AI can step into a meeting with useful information without being asked. In PowerPoint, it can follow your organisation's own brand rules. And across Microsoft 365, agents are becoming a much bigger part of how work gets done.

But more autonomy needs more control. So, at the same time, Microsoft is introducing new ways to govern agents, protect sensitive information and decide when AI should – and shouldn't – be involved.

Here's what caught our attention this month.

Copilot can now speak up in Teams meetings

What's new?

Microsoft Teams Facilitator is becoming a more active participant in meetings.

If an unanswered factual question comes up during a meeting, Facilitator can recognise it, search the web and post a response into the meeting chat – without someone having to stop the conversation and ask Copilot.

It's designed to step in selectively rather than responding to everything it hears, and Facilitator needs to be enabled for the meeting.

It's a small change with a bigger implication. Until now, we've largely thought about Copilot as something you ask for help. Microsoft is starting to explore what happens when AI recognises an opportunity to help for itself.

What does it mean for you?

Think about how often someone says, "Does anyone know when that changed?" or "Can somebody check that?"

Instead of somebody disappearing into a browser while everyone waits, Facilitator could find the answer while the conversation continues.

Useful? Absolutely.

But it's also another reason businesses need clear guidance around AI in meetings. People should understand when Facilitator is active and remember that an AI-generated answer still needs the same common sense and scrutiny as any other information brought into the room.

PowerPoint Copilot is learning your brand

What's new?

Anyone who's asked AI to create a presentation will recognise the problem: it can produce something polished, but that doesn't necessarily mean it looks anything like you.

Microsoft is addressing that with Brand Kit and Skills in Copilot for PowerPoint.

Organisations can create reusable guidance that tells Copilot how presentations should look and be structured. That can include more than colours and logos – it can also guide how your organisation approaches presentations.

Once those skills have been created and published, eligible employees can use them when creating presentations with Copilot.

What does it mean for you?

This could make Copilot considerably more useful for businesses using it to create content at scale. Rather than expecting every employee to remember the right prompt, some of that knowledge can be built into the experience.

That means more consistency, less time fixing AI-generated slides and fewer presentations that technically contain your logo but still don't really look or feel like your business.

There is some set-up involved. Someone still needs to define, test and maintain the rules Copilot follows.

But that's arguably the interesting part. We're moving from "everyone experiment with Copilot" towards businesses deciding how they actually want Copilot to work for them.

You can now tell AI to leave the meeting

What's new?

Not every part of every meeting needs AI listening in.

Microsoft Teams now gives meeting organisers and presenters greater control over Meeting AI during a live meeting.

AI features such as Copilot, Facilitator and intelligent recap can be turned off when the conversation moves into something more sensitive, subject to the organisation's existing policies and licensing.

What does it mean for you?

This is a deceptively useful change.

A normal project meeting could suddenly move onto employee performance, a confidential commercial discussion or something else you don't want included in AI-generated meeting content.

Previously, organisations largely had to decide how AI would be used before the meeting began. Now there's more flexibility to respond to what actually happens in the room.

The technology is simple. The bigger question is whether your people know when they should use it.

If your organisation uses AI in meetings regularly, it's worth making this part of your internal Copilot guidance.

Teams’ meetings are getting a reshuffle

What's changing?

If you've ever clicked the wrong button while presenting in Teams, Microsoft has some good news.

Meeting controls are being reorganised, with microphone, camera and sharing controls grouped more logically and the Leave button separated from other actions.

Users will also be able to customise some controls, while less frequently used options are being moved into a tidier More menu.

The sharing experience is changing too. Microsoft is introducing clearer previews of screens and windows and an additional confirmation before content is actually shared.

What does it mean for you?

Mostly, fewer little Teams frustrations.

But the sharing change is worth highlighting.

Accidentally presenting the wrong window isn't just embarrassing. It can expose emails, customer information, internal messages or other information that wasn't meant for everyone on the call.

An extra opportunity to check "Yes, that's definitely the screen I want" is a small but welcome improvement.

There's no major action required. Just don't be surprised when familiar controls start appearing in slightly different places.

Microsoft is putting more guardrails around AI

What's changing?

As Copilot becomes capable of doing more, Microsoft is also tightening the controls around what AI can access and create.

There are several changes worth knowing about.

Agents created through Agent Builder can be submitted for wider use within an organisation, but administrators can control approval before they're made available through the organisation's Agent Store.

Management of Federated Copilot Connectors is also becoming more centralised, giving IT teams greater oversight of the external information sources being connected to Copilot.

And when Copilot creates a file using information with sensitivity labels, Microsoft can carry the highest applicable sensitivity label through to the newly generated file.

What does it mean for you?

This is the less glamorous side of Copilot, but potentially the more important one.

It's one thing when an employee uses AI to rewrite an email.

It's another when AI agents can access company information, connect to other systems, create new files and carry out work.

Businesses need to know which agents are being created, what they can access, who can use them and who's responsible for them.

Microsoft is increasingly providing the controls to do that.

The important bit is making sure someone in your organisation is actually using them.

Data protection is stretching beyond Microsoft 365

What's new?

Your sensitive information doesn't conveniently stay inside Microsoft.

That's why Microsoft is expanding Purview Data Loss Prevention capabilities to help protect sensitive information in connected third-party cloud services, including platforms such as Google Workspace and Box.

Microsoft is also introducing simulation capabilities for automatic sensitivity labelling, allowing administrators to see what a policy would label before enforcing it.

What does it mean for you?

For businesses working across several cloud platforms, this is an important direction of travel.

Moving a document out of SharePoint doesn't suddenly make the information inside it less sensitive. Extending protection beyond the Microsoft environment can help businesses apply more consistent rules to their data.

Simulation is useful too.

Rather than switching on an automated labelling policy and hoping you've configured it correctly, IT teams can see the likely impact first and make adjustments before it starts affecting users.

In security, knowing what something will do before you switch it on is rarely a bad thing.

On your radar: Power Apps pricing

Microsoft has announced a price increase for Power Apps Premium from 1 January 2027.

Power Apps Premium will increase by 10% at Microsoft's US list price, while the discounted Premium plan for larger deployments will see a bigger increase. UK pricing can vary depending on your agreement, licensing route and currency adjustments.

Existing subscriptions will retain their current pricing until the end of their existing term.

There's nothing you need to do today, but if you're using Power Apps Premium or have a renewal approaching, it's worth factoring the change into your 2027 budget.

Focus Group customers can speak to us about how the changes could affect their Microsoft licensing costs.

That's September

The most interesting thing this month isn't any individual feature.

It's the pattern.

Microsoft is steadily giving Copilot more context and more freedom to act – while simultaneously building more controls around where it can operate, what it can access and what happens to the information it creates. For businesses, the next stage of Copilot isn't simply about switching features on.

It's about deciding where you want AI to help, where you don't, and putting the right controls around both.

Need help getting more from Microsoft 365, Copilot or your wider Microsoft environment?