Table of Contents
Claude Meetings: what it does, what it costs, and the guest rule that shuts it out
Anthropic shipped a version of Claude that sits in your Google Meet call as a named tile next to yours, not a bot quietly transcribing in the background. It posts a notice announcing itself, drafts the plan the room is discussing while people are still talking, and saves the output to claude.ai the moment the call ends per TechCrunch. Anthropic calls it Claude Meetings, and it launched in beta on November 10, 2026, alongside two companion features, Claude Whiteboards and Claude Sheets.
The feature itself is not the interesting part. Every productivity vendor has a meeting bot now. The interesting part is the one guardrail Anthropic built in on day one: Claude automatically leaves the second an external guest joins, unless a workspace admin explicitly says otherwise. That single rule tells you more about who this is built for than the feature list does.
What actually happens on the call
Claude joins as a visible participant, identifies itself in the meeting chat, and then works in parallel with the conversation instead of after it per TechCrunch. While people talk, it can draft the plan being discussed, map a timeline onto a whiteboard, answer questions typed into the meeting chat, and pull information from connected work apps, Anthropic's own example being last year's figures showing up mid-conversation instead of someone promising to "follow up with that number." Claude Whiteboards runs underneath it, so the board fills in live rather than getting built from a transcript afterward.
Picture a Tuesday roadmap sync where three people are arguing about a Q1 date. Normally someone volunteers to write up the decision after the call, and that writeup drifts from what was actually said. With Claude Meetings running, the draft plan and the timeline exist before anyone hangs up, sitting in claude.ai with the rest of the session's output already attached. Whether that draft matches what the room actually agreed to is the part nobody outside Anthropic has verified yet; TechCrunch's report is based on Anthropic's own demo and product copy, not an outside test.
Who gets it, and what it costs
Claude Meetings is beta, and it is gated to paid workspace tiers.
| Detail | What's confirmed | Source |
|---|---|---|
| Eligible plans | Team and Enterprise only, not Pro or Free | TechCrunch |
| Supported call platform | Google Meet at launch; more platforms "expected to follow" | TechCrunch |
| Extra cost | None reported beyond the existing Team / Enterprise seat price | TechCrunch |
| Status | Beta, not generally available | TechCrunch |
If you're on Claude Pro at the personal level, this doesn't apply to you yet. It's a workspace feature, routed through a workspace admin, built for a team that already pays for Team or Enterprise seats.
How to turn it on
Anthropic hasn't published a step-by-step admin doc for this yet, which tracks for a feature that's one day old. What's confirmed is the shape of the rollout: it's a Team or Enterprise workspace capability, Google Meet only, and the external-guest behavior below is on by default, not something an individual user configures per call. If you run a Claude workspace, the practical move right now is to check your admin console for a Claude Meetings or connected-apps toggle and watch Anthropic's support site for the setup doc, rather than assume a specific click path that hasn't been documented publicly.
The guardrail: why it leaves the moment a guest shows up
Here's the part that matters more than the whiteboard. Claude won't join a call with an external guest unless a workspace admin has authorized it, and if a guest joins a call Claude is already in, it leaves automatically per TechCrunch. No manual removal, no setting to override mid-call. The default is exclusion, and someone with admin rights has to flip it the other way in advance.
That default tells you Anthropic is worried about the same thing any security-literate buyer would ask first: what happens to a client call, a vendor negotiation, or a board meeting where a non-employee is on the line. An AI participant with access to connected work apps and a live chat channel is a data-exposure risk the instant someone outside the company can see what it says or does. Building the opt-out as "leaves by default" rather than "stays unless removed" is the correct call, and it's also the reason this ships to Team and Enterprise first: a feature with an admin-authorization gate assumes there's an admin.
What it does not cover yet: Zoom and Microsoft Teams, where a lot of external-facing meetings actually happen. Anthropic says more platforms are coming, with no date attached. Until then, Claude Meetings only helps the internal half of your calendar.
Claude Meetings versus the meeting-notes tools you already have
Granola, Otter, and Fireflies have been in this space for years, but they solve a different problem. They listen and summarize after the fact. Claude Meetings works during the call and is visible to everyone on it.
| Tool | Joins as a visible participant | Acts during the call | Where it runs | Plans |
|---|---|---|---|---|
| Claude Meetings | Yes, named tile, posts a chat notice | Yes, drafts plans, builds a whiteboard, answers chat | Google Meet (beta) | Team / Enterprise only |
| Granola | No, runs alongside, not a call participant | No, notes and summary after | Zoom, Meet, Teams | Free tier + paid plans |
| Otter | No | No, transcript and summary after | Zoom, Meet, Teams | Free tier + paid plans |
| Fireflies | No | No, transcript and summary after | Zoom, Meet, Teams | Free tier + paid plans |
Plan and platform details per each vendor's current pricing page; Claude Meetings details per TechCrunch, Nov 10, 2026.
If your team's complaint is "nobody writes up what we decided," any of the four fixes it. If the complaint is "we keep re-explaining context Claude already has from our docs and calendar," only Claude Meetings closes that gap, because it's the only one of the four that can act on connected apps mid-call instead of just listening to it.
Who should turn this on
If you're already paying for Claude Team or Enterprise and most of your meetings are internal, this is close to free upside: turn it on, let it run in your next planning sync, and check whether the drafted plan actually matches what the room agreed to before you trust it unsupervised. If your calendar is mostly external, client calls, sales demos, vendor syncs, this does nothing for you yet; it stays off those calls by design, and Zoom and Teams support isn't here. And if you're not on a workspace plan at all, the decision isn't yours to make. It's your Claude admin's, and the next real signal to watch for is whether Anthropic's "Project Parka" meeting-recorder rumors turn into a second, broader version of this that works without an admin's authorization gate standing in the way.