Anthropic is taking its assistant from the chat window into the meeting itself. The company announced the beta launch of Claude Meetings on November 12, allowing eligible organizations to add Claude as a visible participant in Google Meet conversations. The assistant can follow discussion, answer questions and create work products while people are still talking.
In its product announcement, Anthropic describes a collaborative workflow in which participants can ask Claude to assemble notes, documents, whiteboards or other artifacts as decisions are being made. That is a different promise from the familiar post-meeting summary: it puts the drafting stage inside the discussion rather than after the call ends.
What teams can actually do
Claude appears as an identifiable participant. Anthropic describes the assistant as a participant that can respond to team questions during a call. The default experience emphasizes written responses rather than autonomous changes. The assistant may use connected tools with appropriate permission and can organize the output on a meeting page for later work.
The official setup guide says the beta is rolling out to Team and Enterprise plans in Claude and Claude Desktop. Google Meet is the initial conferencing platform. Anthropic says that eligible users add Claude to calendar-linked meetings, while organizations control availability through administrative settings.
On Team plans, the feature is enabled by default subject to an owner's ability to turn it off. On Enterprise, an owner must enable access, either broadly or for selected roles. This matters because inviting an assistant is not simply a personal productivity choice when a call includes customers, employees or confidential business information.
Important restrictions and consent
Anthropic says the beta is unavailable for some organizations using zero-data-retention configurations, HIPAA-ready setups or customer-managed encryption keys. These are meaningful eligibility limits, not evidence that every other call is automatically suitable for AI transcription. The company's help page explicitly instructs users to ensure everyone agrees before a meeting is transcribed.
Organizations also need rules for what information Claude can access and what it may create. A meeting with sensitive material could expose data to downstream documents or connected applications if privileges and retention expectations are not clear. Meeting hosts should distinguish joining the call, hearing discussion, writing notes and taking actions in external systems: those are different permissions.
Why this is more than a notes feature
Many conferencing platforms already produce summaries and action lists. Claude Meetings shifts competition toward real-time participation: turning an evolving discussion into editable work before participants disconnect. That may reduce administrative follow-up, but it can also encourage teams to accept plausible-looking drafts without checking details.
For users, the immediate test is practical. Does Claude accurately separate decisions from unresolved ideas? Does it flag uncertain statements instead of turning them into commitments? And can owners audit who invited it and which external tools it used? Anthropic's launch explains the advertised features and availability, not independent evidence of accuracy or productivity improvements.
The rollout remains a beta, so businesses should start with low-sensitivity meetings, obtain participant consent and verify resulting documents before relying on them. The significance is not that every meeting now needs an AI attendee, but that AI tools are beginning to operate within the shared workspace where decisions are actually made.
Sources
Anthropic launch announcement, November 12, 2026; Anthropic Help Center, Claude Meetings setup and eligibility.