Anthropic’s Claude Help Center now documents Claude Meetings, a beta that sends Claude into Google Meet calls as a named participant. Once inside, it writes notes, fields questions from attendees, and drafts docs, whiteboards and slides while people are still talking. AI Insiders reported on 11 November that Meetings was only marked “coming soon”; this page turns the promise into a feature with settings, rules and a bill.
Availability depends on the plan. Meetings is on by default for Team plans, and an owner can switch it off. On Enterprise, nothing happens until an owner enables it, either for the whole organization or for part of it through custom roles.
Claude enters a call in one of two ways: a person adds it, or that person has auto-join turned on under Settings, Capabilities, Meetings. Auto-join can cover every meeting or every meeting except one-on-ones. Owners cannot set it for someone else. If the host has to approve entry and refuses, Claude stays out of that call until somebody adds it again, and the person who invited it sees that it was turned away.
The output lands on a meeting page. It holds a summary, action items, the full transcript and anything Claude built during the call. Colleagues from the organization on the invite receive a recap email and can edit Claude’s documents, and the page can be shared more widely inside the company. A calendar view lists past and upcoming meetings, and in an ordinary chat a user can later question Claude about earlier meetings they are allowed to see.
Outside guests are the sharpest edge. Team plans let Claude join those calls unless an owner blocks it, while Enterprise owners must opt in first. With guests blocked, Claude does not enter a call that has an external invitee, and it leaves if one arrives midway, another company’s notetaker bot included. With guests allowed, Claude listens and takes notes but posts only its notice in the call chat and never replies there. Guest remarks still appear in the notes and transcript, yet guests receive neither the meeting page nor the recap email. Each user also gets a personal setting for external-guest meetings.
On privacy, Claude announces itself in the call chat as it arrives. Anthropic says it stores a transcript, a summary and descriptions of what was on screen, and does not keep audio. The help page offers no retention period and no detail on who inside Anthropic can read the stored text, questions a compliance team will want answered before approving it.
Pricing has two tiers. Notes, transcripts and answers that avoid connected tools come with Team and Enterprise plans. Docs and whiteboards made in meetings are free through 10 February 2027, and the page names no price for afterward. Anything that calls connected tools counts against the user’s own usage and can generate charges for the organization, so a meeting where Claude pulls from a CRM is billed differently from one where it only listens.
The page also marks its limits. It describes Google Meet only and names no other video platform, so teams on Zoom or Teams have nothing here to plan around. It offers no accuracy figures for notes or transcripts.
Operators have about three months before the free window on live docs and whiteboards closes. Pilot it on internal calls now, decide the external-guest policy before an owner flips the switch, and learn the real cost of connected tools before February.
Anthropic’s Claude Help Center, “Use Claude Meetings,” support article read 13 November 2026.
