Anthropic Puts Claude Inside Video Calls as a Visible Meeting Participant

Anthropic began rolling out Claude Meetings in beta on November 12, 2026, and the feature puts an AI participant directly into video calls on Team and Enterprise plans. Claude answers spoken and written questions, drafts documents during the discussion and leaves every meeting with a persistent page containing notes, a transcript and its outputs.

For HR and workplace technology teams, the launch moves AI note-taking out of post-meeting tools and into the live conversation. It also lands consent, guest policy and administrator control squarely on the desk of whoever governs meeting tools. Enterprise administrators must enable the feature deliberately, for all or part of their organizations. Team plans have meeting invitations on by default, and administrators can disable them.

What can Claude actually do in a call?

Participants can address Claude aloud or in chat. It replies in writing by default and can call on connected tools when the relevant person grants permission. It builds documents, whiteboards and other artifacts as the conversation develops, while teammates on the invitation refine them in real time.

Anthropic's own example shows the pattern. A team planning a customer event asks Claude to draft a plan from what has already been said. When the budget comes up, Claude requests approval to retrieve the previous year's costs, posts the figure in chat and compares it against the current plan. The group moves the timeline onto a whiteboard while still editing the document, then asks Claude for a recap before the call ends.

Each meeting gets a separate claude.ai page holding its notes, transcript and everything Claude produced. Invited staff can view and edit Claude's outputs and share the page within the organization. Attendees can return to past meetings, ask Claude about them or schedule Claude into future meetings from their claude.ai calendar.

What happens to privacy and guest access?

Anthropic has drawn the consent boundary explicitly. Claude appears as a visible participant and posts a chat notice when it joins. It saves a transcript, a summary and screen descriptions rather than audio. Anthropic tells users to confirm that everyone on a call agrees to be transcribed before adding Claude or enabling auto-join — language that procurement and legal teams will read closely.

External guests change the defaults. Claude stays out of a call when someone outside the organization is invited, and it leaves if an outsider joins mid-call. An administrator can override this. On permitted external calls, Claude posts its notice but does not reply in chat, while guest speech still appears in the notes and transcript — a distinction worth flagging to anyone handling confidential candidate or client conversations.

Connected tools run under each individual's permissions. Approval is requested once and then carries forward. For spoken requests, Claude currently asks each time to confirm who is speaking. Anything Claude posts in call chat is visible to everyone on the call.

What are the rollout limits and costs?

Google Meet is the only supported platform at launch. Anthropic describes other video-conferencing platforms as coming soon. Transcripts will also flow to Gong, Attention, Granola, Fireflies.ai and Otter, though Anthropic lists those integrations as planned rather than part of the beta.

The beta excludes several compliance-sensitive configurations: organizations using zero data retention, a HIPAA-ready configuration or customer-managed encryption keys cannot use it at all. That exclusion effectively tells regulated employers where the feature stands today.

Pricing during the beta is simple. Notes, transcripts and in-call answers are free on Team and Enterprise plans, unless an answer uses a connected tool, in which case it counts toward plan usage. Documents, whiteboards and other artifacts created during calls stay free through February 10, 2027.

What should employers watch next?

The capability questions are concrete rather than speculative. Which meetings justify a transcribing participant, and who decides when external guests are present? How do administrators audit connected-tool permissions once approvals carry forward? And will the promised platform and transcript-tool integrations arrive before the free artifact window closes in February 2027?

Claude Meetings shifts drafting into the call and the call into a managed workspace. Its practical fate in most organizations will rest on transcription consent, guest policy and connected-tool permissions as much as on Claude's ability to write the first draft.