Claude as a participant, not a recorder
Assistants that transcribe meetings have existed for years. Copilot in Teams, Gemini in Meet, Otter, Granola, Fireflies. Claude Meetings enters a mature market, as Gartner noted, and the reason it is worth a look is not transcription.
It is that Claude does more than listen. It joins the call as a visible participant and works while people talk. If you agree on a plan during the meeting, it can write it. If you need a figure, it can look it up in connected tools. If you are working through a sequence of tasks, it can draw it on a shared whiteboard that everyone sees and corrects.
Anthropic presented it on November 10 together with Claude Sheets, Whiteboards and Data, and the beta started on November 12. Craig Durr, an analyst at The Collab Collective, summed it up well: Claude moves from a place you go to ask a question to a place where teams do their work.
How it works in practice
You invite Claude to a meeting from the calendar inside claude.ai, or turn on automatic joining. When it joins, it shows up among the participants and posts a notice in the chat.
During the call you can ask it questions aloud or in chat, and by default it answers in writing in the meeting chat. It can use each person's connected tools, with their permission: it asks the first time, and the answer holds from then on. For spoken requests it asks each time to confirm who is speaking. Before posting something from a tool in chat, a budget line for example, it asks for approval. And remember that whatever it writes in chat is visible to every participant.
At the end of the meeting you can ask for a recap to share. Either way, each meeting gets a page in claude.ai with notes, assigned action items, the transcript and everything Claude built during the call. Invited internal participants can view and edit it, and it can be shared within the organization. Later you can go back to past meetings and ask about what was decided.
Platforms, plans and costs
At launch Claude Meetings works only with Google Meet. The official page says support for other platforms is coming soon, with no dates and no names. An Anthropic spokesperson also mentioned Microsoft Teams to Computerworld, but until it shows up in the documentation I treat it as an announcement, not availability. For the many companies that meet on Teams, that is the main limit today.
The beta is reserved for Team and Enterprise plans. On Team, the ability to invite Claude is on by default and the admin can turn it off. On Enterprise the admin turns it on, for the whole organization or part of it.
On costs Anthropic was precise. During the beta notes and transcripts are free. So are answers in meetings, except when they use a connected tool: then they count toward the plan's usage limits. Documents, whiteboards and other artifacts created in meetings are free until February 10, 2027. What happens after that has not been said.
Get updates on Claude and AI for business
One email when there's something worth reading. No spam.
Evaluating Claude for your organization? Learn what it costs or which plan fits best
What it keeps, and who is left out
Claude saves the transcript, a summary and descriptions of what is shown on the shared screen. It does not keep the audio.
On external guests the default is cautious. Claude does not join meetings with invitees from outside the organization, and if an outsider joins once the meeting has started, it leaves. The admin can change this setting. In that case Claude still posts the notice in chat but does not reply there, and what guests say ends up in the notes and transcript.
Some organizations are excluded from the beta: those with zero data retention, the HIPAA-ready configuration for US healthcare rules, or customer-managed encryption keys. If your company uses one of these configurations, Claude Meetings is not an option for now.
Soon, according to Anthropic, transcripts can be sent to tools like Gong, Granola, Fireflies.ai, Otter and Plaud, useful for anyone who already has a sales process built on one of them.
The European angle: transcribing meetings is not neutral
Anthropic itself advises checking that everyone is comfortable with transcription before inviting Claude or turning on automatic joining. The analysts quoted by Computerworld add that recording and consent rules vary by country and by company, and that IT, legal and HR should agree on when Claude may join.
In Europe this has concrete implications. Under the GDPR, transcribing a meeting is processing of personal data, so you need a legal basis and a notice to participants. Many countries also have specific rules on tools that allow employers to monitor employees' activity. In Italy, for example, article 4 of the Workers' Statute governs tools from which remote monitoring of employees may result, and a tool that transcribes and summarizes internal meetings should be assessed under that lens too, together with whoever handles privacy and labor relations in the company.
I'm not writing this to discourage anyone. I'm writing it because on Team the feature is on by default, and the concrete risk is that someone invites Claude to a meeting before the company has decided the rules. On approaching privacy more generally, see Claude and GDPR for businesses.
My take
Claude Meetings is the product in the November bundle that interests me most, and the one I'd be most careful about putting into production.
It interests me because the meeting is where work gets decided and then lost. The value is not in the minutes, which have existed for years. It is in the plan decided at eleven being written, shared and linked to the right tools by half past eleven, instead of becoming an email someone sends the next day.
I'm cautious for three reasons. It only works on Google Meet. It is in beta, and Gartner is right to say a broad work surface is not, on its own, proof of enterprise value. And above all Nicholas Lin, a product lead at Anthropic, told Axios that today Claude acts only when asked, but could become more proactive over time. A participant that acts on its own opens a different governance conversation.
What I'd do before inviting it
If you have a Team plan, the first decision is whether to leave the feature on. If your company doesn't yet have a rule on transcribing meetings, I'd turn it off until it does.
Then I'd write the rule, briefly: which meetings Claude may join, who decides, how participants are informed, who sees the notes and how long they are kept. With your privacy lead and, where needed, with HR.
Finally I'd run a real test with a small group that already uses Google Meet: the weekly meetings of a project team, for example. After a month the questions are simple. Are agreed actions carried out more often? Is less time lost reconstructing who was supposed to do what? If the answer is yes, the rule is ready to be extended. If it's no, you found out cheaply.