On November 12, 2026, Anthropic announced on its official blog that Claude Meetings is entering beta: Claude can join video calls as a visible participant and get to work while the meeting runs — answering questions, drafting documents, and building whiteboards during the call, then automatically organizing meeting notes and to-dos afterward. It is rolling out first to Team and Enterprise plans, starting with Google Meet.
What can it actually do in a meeting?
Using it is simple: someone adds Claude to a meeting, or turns on auto-join, and it shows up in the participant list and posts a join notice in the chat. During the meeting you can ask it questions out loud or type them in the chat, and it answers in writing by default. Anthropic's example is planning a customer event: someone says, "Claude, start writing the event plan," and it drafts a document from what has already been discussed while colleagues keep editing as they talk; when the budget comes up, someone asks what last year's event cost, it asks permission before looking it up, then posts the figure in the chat next to this year's plan; when the timeline comes up, someone asks for a whiteboard and the team schedules the work on it together; before the meeting even ends, it can send the notes to the whole group. This follows the same pattern as the whiteboards and sheets Claude launched on November 10: produce something usable inside the meeting room instead of sorting it out afterward.
Outsiders in the call, recordings saved — where are the boundaries?
Privacy gets the most space in the announcement. Claude creates a separate page for each meeting, holding the notes, transcript, and what was produced, which invited people in the organization can view and edit; by default, Claude will not join if someone from outside the organization is invited, and it leaves automatically if an outsider joins midway — an admin can lift that restriction. On calls with outsiders, Claude only posts a notice in the chat and does not reply there, though what outsiders say still goes into the notes and transcript. Audio is not saved; what is kept is the transcript, a summary, and screen descriptions. When it calls someone's connected tools, it acts under each person's own permissions and asks first the first time; for spoken questions, it confirms who is speaking each time. The beta is not open to organizations with zero data retention, HIPAA-ready configurations, or customer-managed encryption keys. On pricing, notes, transcripts, and in-meeting Q&A are free on Team and Enterprise during the beta, and documents, whiteboards, and other artifacts are free through February 10, 2027.
The meeting-AI field is already crowded — Microsoft Copilot, Google Gemini, Otter, and Fireflies all do notes and summaries. Claude's difference is that it does more than record: it works hands-on in the meeting, using each person's tools. Two things are worth watching: when other meeting platforms get connected, and what it costs once the beta ends. Anthropic also says it is working with Gong, Attention, Granola, Fireflies.ai, and Otter so transcripts can later be sent to tools teams already use. For teams that run meetings across time zones and still have someone write up the notes afterward, the value needs no waiting: some people have already stopped taking meeting notes.
