What can Microsoft 365 Copilot do in Teams meetings?

Microsoft 365 Copilot in Teams can summarise key discussion points, identify who said what, suggest action items and answer questions about a meeting. It can reason over both spoken discussion and meeting chat, helping users catch up, identify disagreements, review feedback and generate follow-up material.

Does Copilot in Teams require transcription?

Transcription is not required to activate Copilot for live meeting insights. However, live transcription is required to ask Copilot questions after the meeting ends or retain its conversation history. If transcription stops, the persistent post-meeting Copilot experience stops too.

How should organisations prepare for Microsoft 365 Copilot in Teams?

Preparation should include reviewing SharePoint permissions and oversharing, sensitivity labels, external sharing, retention policies, conditional access, audit logging and meeting transcription settings. Copilot respects existing permissions, so organisations should address access-control weaknesses before deployment.

Can Copilot in Teams work in external meetings?

Microsoft states that Copilot will not work in meetings hosted outside the participant's organisation. This is an important consideration for organisations whose collaboration relies heavily on customer, partner or supplier meetings hosted in external tenants.

How should organisations measure Microsoft Copilot in Teams ROI?

Measure more than active use. Track meeting follow-up time, clarity and completion of actions, time required for late attendees to get context, repeat-meeting duration, decision-making speed and role-specific user feedback. The strongest evidence links time saved to a meaningful operational outcome.

Is intelligent recap included with Microsoft 365 Copilot?

Yes. Microsoft says intelligent recap is included with a Microsoft 365 Copilot licence. It is also available through Teams Premium, subject to the organisation's licensing and configuration.

Front page

Case StudyProductivity AI22 Jul 2026 · 10 min read

Microsoft 365 Copilot in Teams: From Licence Purchase to Everyday Collaboration Tool

Deploying Microsoft 365 Copilot in Teams? Learn the governance, data and adoption steps that turn AI features into collaboration value.

Alex Cole

Microsoft 365 Copilot in Teams licence to daily use

Enterprise collaboration leaders do not need another explanation of what an AI meeting recap is. They need to know what happens after they buy the licences: which Teams policies must change, where sensitive data becomes a blocker, how to move users from curiosity to habit, and how to prove that Copilot is reducing work rather than creating another layer of it.

Elaine Smith, Program Lead for Copilot, PepsiCo, expressed her opinion on using Copilot.

“By consolidating on Microsoft 365, we’ve created an environment where Copilot can do its best work. This is probably why we’re seeing 90% to 95% daily active use for Copilot.”

That experience at PepsiCo is instructive precisely because it did not begin with Copilot. The company first standardised its collaboration estate on Microsoft Teams, then deployed Microsoft 365 Copilot across an organisation of more than 320,000 employees in 200 countries. It also migrated 3,300 video rooms to Teams Rooms. Microsoft reports that PepsiCo now has 90% to 95% daily active use for Copilot among its users, although that figure is vendor-published and should be read as customer evidence rather than an independently audited benchmark.

Related Articles

TL;DR — The Deployment Reality

Copilot in Teams is a data-and-governance deployment, not just an AI rollout. Permissions, transcript retention, sensitivity labels and content hygiene shape the experience users receive.
Start with a narrow set of high-frequency workflows: meeting catch-up, action tracking, chat synthesis and call follow-up.
Persistent post-meeting value depends on transcription. Without a transcript, Copilot can work live but cannot retain meeting context after the session.
PepsiCo's reported 90% to 95% daily active use followed standardisation and structured change management — not a licence-only deployment.
The strongest fit is a mature Microsoft 365 estate. Organisations with fragmented collaboration data, loose permissions or mixed meeting platforms should resolve those foundations first.

What Does Microsoft 365 Copilot in Teams Actually Change?

Direct answer: Microsoft 365 Copilot in Teams applies generative AI to meeting discussion, calls, chat and channels. Its distinguishing proposition is not simply AI summaries, but its ability to reason across collaboration content already held in the Microsoft 365 environment — subject to the permissions and policies governing that content.

In a Teams meeting, Copilot can identify discussion points, attribute comments, suggest actions and answer questions about what happened. Users can ask where people disagreed, what a particular participant said, or which issues remain unresolved. It can also use meeting chat alongside the transcript and export generated material to Word or Excel where organisational policy allows.

The broader deployment opportunity is cumulative. A user can move from a Teams meeting recap to a related file, chat history or email context without leaving the Microsoft 365 environment. For organisations that already use Teams as their primary collaboration layer, that is a meaningful advantage over standalone AI note-taking tools that can document an interaction but cannot necessarily connect it to the work that followed.

Microsoft positions Copilot across Teams meetings, calls, chats, channels, Teams Rooms and Microsoft Places. It also offers AI-enabled capabilities including Facilitator, Channel Agent and Interpreter. The temptation will be to treat this as one large deployment. That would be a mistake. The practical route is to establish a few repeatable collaboration use cases before introducing broader agent, calling and workplace scenarios.

Buyer Lens

Best initial use case: helping late joiners catch up, producing decision/action summaries and reducing post-meeting administration.
Core differentiator: Microsoft Graph context across meetings, chats, files and email — where the organisation has structured its content and access well.
Core limitation: Copilot only works as well as the collaboration data, permissions and retention rules it can safely access.

Before the Pilot: Data Readiness Is the Real First Step

Direct answer: Before assigning Microsoft 365 Copilot licences, IT teams should review SharePoint permissions, external-sharing settings, sensitivity labels, retention, conditional access and audit logging. Copilot respects existing permissions — but it can make previously overlooked access problems easier for users to discover.

This is the part of the deployment that gets missed when Copilot is treated as an app rather than a service layered across the tenant. Microsoft recommends that organisations assess oversharing in SharePoint, review popular sites, conduct access reviews, limit broad permissions and consider restricted SharePoint search for sensitive content. It also recommends multifactor authentication, conditional access and unified audit logging as part of its readiness guidance.

That does not mean Copilot creates a new permission model. It does mean it exposes the consequences of the one an organisation already has. A loosely governed SharePoint estate may have remained manageable when employees had to know where to look. An AI assistant that can surface relevant information through natural-language prompts changes that discovery dynamic.

Ashok Paranjothi, VP of Global Workplace Services, PepsiCo also shared positive views:

“Security and responsible AI is top of mind for us at all times… Copilot is fully built on the Microsoft 365 access control. That’s made life much, much easier and simple for us in rolling out the capability for our associates.”

The distinction matters. Existing access controls are an advantage for Microsoft-centric enterprises, but they are not a substitute for data governance. Buyers should ask: who can see what today, which content has no clear owner, how are sensitivity labels applied, and would an employee receiving a well-formed answer from Copilot expose a weakness in our permission design?

Transcription Policy Is Where the Collaboration Experience Becomes Real

Meeting AI is not a single on/off switch. Teams organisers can set Copilot and Facilitator to operate during and after a meeting, during the meeting only, or not at all. Those decisions are shaped by organisational policy, sensitivity labels and meeting templates.

The most important implication is persistence. Copilot can be activated in a meeting without transcription, using speech-to-text for live insights. But live transcription is required if users want to ask questions after the meeting or retain Copilot conversation history. Stop transcription and the persistent post-meeting experience stops too.

That forces an explicit policy decision. A legal, HR or executive meeting may need Copilot disabled entirely. Other meetings may permit live assistance but not a retained record. Routine project meetings may benefit from transcription, recap and follow-up actions by default. A credible deployment defines these scenarios upfront rather than presenting every user with the same ambiguous choice.

There are further operational constraints. Copilot does not work in meetings hosted outside a participant's organisation, and deleting a meeting transcript removes Copilot availability and prior conversation history. Sensitivity labels and settings that prevent copying or forwarding meeting content can also block exports of Copilot responses to Word or Excel. These are sensible guardrails, but they illustrate why the user experience cannot be separated from compliance configuration.

How Should Organisations Pilot Copilot in Teams?

Direct answer: Start with a cross-functional pilot group, limited meeting scenarios and measurable collaboration problems. Expand after validating governance settings, user behaviours and evidence of time or coordination savings — not simply after measuring logins.

Microsoft's own deployment model follows three phases: pilot, deploy and operate. It recommends early adopters and champions from different business groups, then a broader rollout supported by adoption reporting and user feedback.

For Teams specifically, a sensible pilot should focus on roles with heavy meeting and follow-up workloads: project managers, sales leaders, operations teams, consultants and distributed leadership groups. Give them a small number of explicit workflows to test:

Catch-up: Can a late attendee recover context faster without interrupting the meeting?
Decisions and actions: Are next steps clearer, correctly attributed and easier to find after the call?
Chat synthesis: Can users summarise a fast-moving channel or prepare for a meeting using the relevant thread?
Follow-up: Does the tool reduce the time spent converting conversation into a document, task list or stakeholder update?

PepsiCo's published experience supports the staged approach. Its adoption team began with simple, recognisable habits: scheduling calls in Teams and using Teams for chat. The company then paired the technology rollout with stakeholder alignment, communications, training and visible executive support. Smith later added:

“We started with small steps, and that’s key when you’re asking so many people to change a daily habit.”

What Should Buyers Measure After Deployment?

Usage is necessary but insufficient. A high active-use figure may show that Copilot is available and attractive; it does not automatically prove that collaboration is better. Teams leaders should pair adoption data with workflow outcomes: time spent producing meeting notes, number of overdue actions, time to onboard a late-arriving stakeholder, repeat-meeting duration, and user confidence that decisions have been captured correctly.

Microsoft points customers toward the Copilot Dashboard in Viva Insights and Microsoft 365 usage reporting to track readiness, usage and sentiment. Those metrics should be supplemented with role-based qualitative feedback. A sales leader, programme manager and compliance officer will experience the same Teams feature set very differently.

The strongest measure is not “hours saved” in isolation. It is whether the reclaimed time changes an outcome: faster decisions, fewer missed handoffs, better client follow-up, less duplicate work or more useful meetings. That is the standard against which a collaboration AI deployment should be judged.

Verdict: A Strong Collaboration Layer — If the Foundations Are Already There

Microsoft 365 Copilot in Teams has a credible advantage for organisations that have made Teams and Microsoft 365 the centre of how work happens. Its ability to connect meeting conversation with chat, documents and wider work context is deeper than a standalone AI meeting assistant can generally offer.

But it is not a shortcut around collaboration discipline. It relies on trusted permissions, usable content, considered transcription rules and a rollout programme that makes new behaviour easy to learn. The product's biggest strength — access to the Microsoft 365 environment — is also why a poorly governed tenant can limit its value or raise risk.

Buyer verdict: Microsoft 365 Copilot in Teams is best suited to organisations with a standardised Microsoft 365 estate that are ready to treat AI collaboration as a change-management and governance programme, not a licence assignment exercise. Start with the meeting workflows people already struggle with. Establish the controls. Measure outcomes. Then expand.

[button_cta link="https://www.uctoday.com/uc-awards/"]Explore the UC Awards[/button_cta]