Microsoft shipped a quiet but consequential update to Microsoft 365 Copilot: the AI assistant can now search your meetings by what was _discussed_ inside them, not just by the subject line on your calendar. The feature reached general availability in June 2026 for Microsoft 365 Copilot customers in the worldwide standard multi-tenant cloud, according to the company’s official roadmap.
What changed under the hood
Until now, finding a past meeting meant hunting through your Outlook calendar or Teams history by title, date, or participant names. Copilot’s new capability lets you describe what happened in the meeting and then retrieves matching sessions. The system pulls from meeting-body content—the text inside the invite—as well as associated Teams chat and the full transcript, if one exists.
Microsoft’s roadmap entry for feature ID 559110 states simply that Copilot can “locate relevant meetings by topic or keyword.” That sounds straightforward, but the practical shift is big. A user who remembers that a budget number, a customer name, or a technical bug was mentioned can now ask Copilot to find the meeting, even if the calendar event was labeled “Weekly Sync” or “Q3 Review.”
The rollout covers the Microsoft 365 Copilot experience across Android, iOS, Mac, Windows desktop clients, and the web. The feature is listed as launched—not in preview or a phased rollout—and the last roadmap update was July 15, 2026.
What the new search means for everyday users
For anyone who lives in meetings, the immediate benefit is less time wasted retracing steps. Instead of scrolling through weeks of Outlook entries or Teams channels, you can ask Copilot a natural-language question like “Find the meeting where we discussed the Acme contract renewal” or “Show me the meeting where we talked about the login bug on the mobile app.”
That’s particularly useful in organizations where meeting titles are placeholders. A “Project update” meeting could cover a dozen distinct topics. Copilot’s semantic search—drawing on AI models that understand the context of words—can surface the right recording even when your memory of the date is fuzzy.
However, the feature’s usefulness depends on what content actually exists. A meeting without a transcript or meaningful chat history gives Copilot very little to search. If your team doesn’t record meetings or if chat threads are sparse, the AI has no magic way to reconstruct what was said. Microsoft hasn’t published details about how Copilot ranks results when multiple meetings touch on the same keyword, nor whether it favors transcript matches over chat matches. Those implementation details will matter as organizations test the feature internally.
What admins should know
The new search capability doesn’t create new data or change access controls. It simply makes existing meeting artifacts—transcripts, chats, invite text—more discoverable through Copilot. That distinction is important for admins who might worry about privacy. A user can only find meetings they already had permission to see. If a meeting chat was private between two colleagues, Copilot won’t expose it to someone else just because they type the right keywords.
Still, the feature sharpens the need for sound information governance. Organizations that have allowed Teams recordings and transcripts to pile up without retention policies now have a tool that proactively surfaces that content. An employee could ask Copilot to “find all meetings that mentioned the compensation review,” and if transcripts exist, they might get results stretching back months or years. Admins should review their Microsoft 365 retention, deletion, and eDiscovery settings before broadly encouraging use of the new search.
Microsoft hasn’t spelled out any new compliance or audit controls for the feature, so the standard Copilot admin tools apply. That includes the ability to control which users have Copilot licenses, whether Copilot can access organizational data, and how the tenant’s data handling policies intersect with AI indexing. The roadmap entry offers no separate toggle for meeting search—it appears to be on by default for licensed users.
How we got here
This is the latest step in Microsoft’s steady march toward making work data more queryable through natural language. Copilot already faced criticism for being an expensive add-on whose value was hard to prove in the first year after launch. Features like meeting search aim to demonstrate the AI’s utility beyond writing emails or summarizing documents—tasks that free tools increasingly handle well.
Meeting intelligence has been a focus for Microsoft for years. Teams Premium introduced intelligent recap in early 2023, using AI to generate meeting notes, tasks, and chapters. Copilot then layered onto that with the ability to answer specific questions about a meeting’s content while the meeting was fresh. Searching across meetings, however, required users to know which meeting they wanted to query. The new capability closes that loop, making the historical corpus of meetings searchable without first picking a specific event.
The roadmap entry doesn’t mention any relationship to Microsoft’s other AI announcements in the first half of 2026. In May, the company teased a separate “Microsoft Scout” agent that acts proactively across apps, though that feature is distinct from Copilot’s meeting search. The two could eventually feed into each other—Scout might notice you’re preparing for a client call and proactively pull up relevant past meetings—but Microsoft hasn’t formally connected the dots.
What to do right now
If your organization uses Microsoft 365 Copilot, the feature is likely already live. Here’s how to get value from it quickly while avoiding pitfalls.
For users:
- Start experimenting with natural-language queries. Include specific terms you remember—project codes, client names, technical jargon. Copilot searches across transcription and chat, so the more unique the phrase, the better.
- Be patient with meetings that lack transcripts. If your team doesn’t routinely record meetings, encourage them to start. That’s the only way Copilot can surface spoken content.
- Remember that chat history is also indexed. Even if a meeting wasn’t transcribed, the pre- or post-meeting chat might contain the keyword you’re looking for.
For administrators:
- Audit your Teams meeting policies. Are transcriptions enabled by default? How long are they retained? If your retention period is short, Copilot’s long-term usefulness erodes.
- Review data governance. The new search might surface sensitive conversations that were never meant to be easily searchable. Consider whether your existing labeling and access policies are sufficient.
- Communicate clearly with employees about what Copilot can and cannot access. The tool respects existing permissions, but users may not understand that. Training reduces confusion and prevents over-sharing.
No action is required to enable the feature—it’s on for all Microsoft 365 Copilot licensees in the standard cloud. Government clouds and specialized tenants may see the feature later, but Microsoft hasn’t published a schedule.
What’s next
Meeting search is a building block, not a destination. The next logical step is Copilot proactively surfacing relevant past meetings when you’re composing an email, preparing a presentation, or joining a new meeting. Microsoft’s broader push toward autonomous agents—like the aforementioned Scout—suggests that Copilot will eventually anticipate what you need to remember rather than waiting for you to ask.
For now, the feature’s success hinges on user adoption and organizational hygiene. Metadata-free search is only as good as the data it indexes. If your Teams environment is a graveyard of unrecorded calls and empty chat threads, Copilot won’t find much. But if your company has been diligently capturing meeting intelligence, this update could make that archive meaningfully more useful.