Claude Can Join Your Google Meet Now. 7 Questions I'd Ask Before Auto-Join

Anthropic has put Claude in the meeting itself. In the Claude Meetings beta, Claude joins a Google Meet call as a visible participant, follows the discussion, answers questions, writes the notes and action items, and can start a doc or a whiteboard while everyone is still talking. That is a different bet from a notetaker that records the call and emails you a summary after you hang up.

I have not put this on a client call yet. Everything below comes from Anthropic's own pages, checked on November 13, 2026: the Claude Meetings announcement and the Claude Help Center article "Use Claude Meetings." Beta features move quickly, so re-read those pages before you roll anything out. We also published a longer walk-through of Claude Meetings features, pricing and privacy this morning, including plan limits and what else Anthropic shipped alongside the beta. This post is the shorter checklist I would run before anyone on a software team turns on auto-join.

What actually happens on the call

Setup runs through Claude, not a Meet add-on you install in the browser. You open Meetings in Claude on the web or in Claude Desktop, connect Google Calendar (an owner has to enable that connector for the organization first), pick a meeting, and click "Add Claude." Claude usually joins without asking. If it does ask, the host can let it in. If the host declines, it does not try that call again unless someone adds it.

Once it is in the call, you talk to it by saying "Claude" and then the question, or by typing in the Google Meet chat. By default it answers in writing in that chat. Anyone on your team can ask it to make a doc or a whiteboard. The Help Center says all artifact types are supported, including slides, and each one is attached to a meeting page in claude.ai. Teammates on the invite can edit those artifacts while the conversation continues.

Claude can also use connected tools during the call. The documented example is sending a Slack message. It uses each person's tools with that person's permission: it asks the first time, and a yes applies going forward. For a request made out loud, it currently asks each time to confirm who is speaking. Approval requests show up in the Meet chat when they can, or on the meeting page if Claude needs to confirm which account you are.

Its tile in the call shows what it is doing. You can pause it by asking it to stop listening, or from the meeting page, and you resume it from the meeting page. To take it out, ask it to leave or choose "Remove Claude." If you add it again, it comes back to the same meeting page. It can stay for up to eight hours and warns you before it leaves.

After the call, that page holds the notes (a summary plus action items), the transcript, and everything it built. People from your organization who were on the invite get a recap email with a link, can see the notes, and can edit what Claude made. The page can be shared with others in the organization. In a regular Claude chat, you can later ask about past meetings you have access to. That is the part I care about as a developer: the meeting becomes context you can query when you write the ticket, instead of a recap that dies in someone's inbox.

The calendar and video-platform limits

Two constraints will decide whether this is usable for your team this month.

First, live support is Google Meet only. Anthropic says other video conferencing platforms are coming soon and has not named them or given dates. If your company standard is Zoom or Microsoft Teams, treat support for those platforms as unannounced. The announcement also says Anthropic is working with partners including Gong, Attention, Granola, Fireflies.ai, Otter, and Plaud so you will "soon be able to send a meeting's transcript to the tools your team already uses." That is a coming-soon statement, not something I can confirm is live today.

Second, joining is tied to Google Calendar inside Claude. An owner enables the connector for the organization, then each person connects their calendar and adds Claude to a meeting, or turns on auto-join. Auto-join lives under Settings, then Capabilities, then Meetings. The choices are all of your meetings, or all of them except one-on-ones. Owners cannot set auto-join for you, which is the right default. It also means the riskiest setting is an individual one, not an org-wide toggle. Settle the questions below before anyone turns it on.

7 questions I'd ask before auto-join

1. Has everyone on the call agreed to be transcribed?

Anthropic is direct about this. The Help Center and the announcement both say: before you add Claude or turn on auto-join, make sure everyone on your calls is okay with being transcribed. When Claude joins, it appears as a participant and posts a notice in the call's chat. That is a useful signal, but it is not the same as consent you collected in advance.

For internal planning, a written team policy plus the in-call notice may be enough. For a client call, a candidate interview, or anyone who did not expect an AI participant, I would get a yes first and put it in the invite. How transcript-only capture is treated under the recording laws that apply to your calls is a question for counsel, not a blog post.

2. What does Claude store, and for how long?

According to Anthropic, Claude saves a transcript, a summary, and screen descriptions, not audio. That is a real difference from notetakers that keep audio or video you can play back. It also means you cannot go back and listen for a word the transcript got wrong.

I could not find a retention period specific to meeting pages. In the pages I reviewed, Anthropic has not said how long those transcripts, summaries, screen descriptions, and artifacts are kept, or how deletion works for a single meeting. Before this holds a backlog refinement or a security review, ask your admin how meeting pages fit the organization's retention settings, and whether you can delete a meeting page the way you delete a chat.

3. Can it join a call that has someone from outside the company?

The defaults differ by plan, and this is easy to get wrong if you only read the high-level "by default Claude stays out" language.

On Team plans, Claude Meetings is on by default and an owner can turn it off. Claude can also join calls with outside guests by default, and an owner can turn that off. On Enterprise plans, an owner has to turn Meetings on, for everyone or for part of the organization through custom roles, and an owner has to allow calls with outside guests. The Help Center lists the Enterprise default for those guest calls as internal meetings only.

When outside guests are not allowed, Claude stays out of a call if someone from outside the organization is invited, and leaves if someone from outside joins partway through, including another company's notetaker bot. If a client brings their own notetaker, Claude may leave mid-call.

Owners and Primary Owners manage this under Organization settings, then Products, then Claude Meetings (beta). A third control, "members' connectors in meetings," decides whether Claude can use people's connected tools during calls at all.

4. What do outside guests actually see?

If an owner allows Claude on calls with outside guests, Claude still listens and takes notes, and it still posts its notice in the call's chat, but it does not reply there. What guests say is included in the notes and transcript. Live docs, whiteboards, and the meeting-page chat stay on that page. Guests do not get the meeting page or the recap email. Each person also has their own "Meetings with external guests" setting.

Two consequences follow. A client will see Claude in the participant list and the notice in chat, but they will not see Claude answering in the Meet chat, and they will not get the doc you asked it to start unless you share it another way, for example by putting it on a screen. And anything Claude does post in the Meet chat on an internal call is visible to everyone on that call, including answers that came from a connected tool. Do not treat the Meet chat as a private side channel.

5. Which tools can it touch, and who approved that?

Connected tools are the difference between a notetaker and something that can create work in your existing stack. They are also the fastest way to turn a casual "Claude, send that to Slack" into a message that leaves the room.

The permission model is per person, not per meeting. A yes applies going forward. Spoken requests get an extra speaker check each time, which reduces the chance that one person borrows a colleague's access by voice. I would still want a short written list of which connectors are allowed in meetings, and I would keep the owner's "members' connectors in meetings" control off until that list exists. The Help Center's example is Slack. Whether a given Jira, Linear, or GitHub connector will create a ticket from an action item depends on what you have connected and on that approval flow. I have not tested that path, and I would not let it create tickets on a client project without a human reviewing each one.

6. Are we even eligible for the beta?

Claude Meetings is rolling out in beta on Team and Enterprise plans, in Claude and Claude Desktop. It is not a Free, Pro, Max, or Ultra feature in the pages I reviewed. During the beta it is also unavailable to some organizations, including those with zero data retention (ZDR), a HIPAA-ready configuration, or customer-managed encryption keys (CMEK). If your company negotiated any of those, plan to wait. "Rolling out" also means an eligible org may not have it yet.

You still need the underlying plan. As of November 13, 2026, Claude's pricing page lists Team Standard seats at USD 20 per seat per month if billed annually, or USD 25 if billed monthly, for teams of 2 to 150. Team Premium seats are listed at USD 100 per seat per month if billed annually, or USD 125 if billed monthly. Enterprise is listed at USD 20 per seat per month, billed annually, with a 20-seat minimum, plus usage billed at API rates. Those are plan prices, not a separate Meetings fee. Check the pricing page before you budget, because published prices change.

7. What will this cost once someone asks it to use a tool?

There is no separate Meetings price in the announcement or the Help Center. During the beta, notes, transcripts, and answers that do not use connected tools are included on Team and Enterprise plans. Answers that use a connected tool count toward that person's usage and, in the Help Center's words, can lead to charges for the organization. Docs, whiteboards, and other artifacts made in a meeting are free through February 10, 2027. Anthropic has not said what happens after that date. I would run a short internal trial with usage visible to an owner, and treat post-February 2027 artifact pricing as unknown.

What I would try first

I would not start on a client requirements call. I would pick one internal planning meeting or a retro, add Claude to that meeting only, and leave auto-join off. The test is simple: does the doc or whiteboard it produces need less cleanup than the notes you already write, and do the action items survive contact with the ticket tracker?

Standups are a weak first test unless they turn into a design discussion. The higher-value case is the call where a decision is made and then lost between hang-up and the ticket tracker. Ask Claude to start a doc when that happens, edit it while people are still on the call, and see whether you can ask about that meeting later from a regular Claude chat.

If that holds up, talk to counsel and to the client about consent, guest visibility, and retention, and only then consider a call with someone outside the company. On those calls, Claude will listen and take notes but will not reply in the Meet chat, and the guest will not get the meeting page.

If most of your work is on Zoom or Teams, keep using whatever notetaker already supports that platform. You can still bring a transcript into Claude later. This beta is worth a look because of the live artifacts and the follow-up chat, not because it is a general meeting bot today.

Would you turn on auto-join for internal planning calls this week, or add Claude to one scheduled meeting at a time until you have seen the notes it writes?