Anthropic Puts Claude in the Meeting: AI Joins Video Calls as a Working Participant

On November 12, 2026, Anthropic began rolling out Claude Meetings in beta, turning Claude into a visible participant on video calls that can answer questions aloud-prompted, take notes and build editable documents while the meeting is still happening. The feature is available now to customers on Team and Enterprise plans, with Google Meet as the only supported video platform at launch.

The release matters because it moves AI assistance from after the call to inside it. Drafting, note-taking and summarizing have traditionally been post-meeting chores. Claude Meetings pulls that work into the live conversation, and parks everything in a managed workspace on claude.ai afterwards. That shifts how teams treat a meeting itself: less a discussion to be remembered, more a document to be co-produced in real time.

What can Claude actually do during a call?

Quite a lot, according to Anthropic's announcement. Participants can address Claude by voice or in chat. By default it replies in writing. It can also use connected tools, provided it has the relevant person's permission, and it can produce documents, whiteboards and other artifacts as the conversation develops. Teammates who were on the invitation can refine those outputs while the call is still running.

Anthropic's own example shows the mechanics. A team planning a customer event asks Claude to draft a plan from what has already been said in the call. When the budget comes up, Claude requests approval to retrieve the previous year's costs, posts the figure in chat, and compares it against the current plan. The group then puts the timeline on a whiteboard, keeps editing the document, and asks Claude for a recap before the meeting ends.

That example captures the pitch: the AI isn't taking minutes, it's doing mid-flight work — pulling data, comparing numbers, scaffolding deliverables — with human approval gates at each step.

What happens after the call ends?

Every meeting gets its own dedicated page on claude.ai. The page holds the meeting's notes, the transcript, and everything Claude produced during the session. People from the organization who were on the invitation can view and edit Claude's outputs, and they can share the page with others inside the organization.

The pages are persistent, not disposable. Participants can return to meetings they attended, ask Claude questions about them later, and add Claude to future meetings directly from their claude.ai calendar. In effect, each call becomes a searchable, editable artifact inside the workspace rather than a calendar entry that fades.

Integrations with existing note-taking and conversation-intelligence tools are coming too. Anthropic said it is working with Gong, Attention, Granola, Fireflies.ai and Otter so users will soon be able to send meeting transcripts to those platforms. That support is planned, not part of the current beta.

How does Anthropic handle privacy and consent?

This is where the beta gets cautious. Claude appears as a named participant and posts a chat notice when it joins a call. It saves a transcript, a summary and screen descriptions — not audio. Anthropic explicitly tells users to confirm that everyone on a call agrees to being transcribed before adding Claude or enabling auto-join.

External guests trigger stricter defaults:

• Claude stays out of a call entirely when someone outside the organization is invited.
• Claude leaves the call if an outside guest joins partway through.
• An administrator can override this and allow Claude on calls with outside guests.
• On permitted guest calls, Claude posts its join notice but does not reply in chat. Guest speech is still included in the notes and transcript.

Connected tools run under each individual's permissions, not blanket access. Claude asks for approval the first time it uses a tool, and that approval applies going forward. For spoken requests, Claude currently asks each time to confirm who is speaking. Any answer posted in the call's chat is visible to everyone on that call.

The design reflects a real constraint: whether this feature gets adopted will depend as much on transcription consent, guest policy and tool-permission settings as on Claude's raw ability to draft documents and take notes.

Which plans get it, and what does it cost?

Google Meet is the supported platform at launch, with Anthropic describing support for other video-conferencing platforms as "coming soon." Claude joins a meeting when someone adds it manually or when auto-join is enabled.

Administrative control differs by plan:

• Team plans: meeting invitations are available by default, and administrators can disable them.
• Enterprise plans: administrators must enable the feature, for the whole organization or parts of it.

Some organizations can't use the beta at all. It is unavailable to those using zero data retention, a HIPAA-ready configuration, or customer-managed encryption keys — the compliance-heavy segments where meeting transcription raises the hardest data questions.

Pricing during the beta is mostly free, with two carve-outs:

• Notes, transcripts and in-call answers are free on Team and Enterprise plans during the beta — unless an answer uses a connected tool, in which case it counts toward plan usage.
• Documents, whiteboards and other artifacts created during calls are free through February 10, 2027.

Why this lands where it does

The meeting-assistant market is crowded. Gong, Attention, Granola, Fireflies.ai and Otter — the very tools Anthropic plans to integrate with — already compete to transcribe, summarize and mine sales and internal calls. Claude Meetings attacks from a different angle: instead of analyzing the conversation afterwards, it participates in it, with tool access, document generation and permissioned data retrieval live on the call.

That's a bet on agentic AI in one of the highest-frequency workplaces settings there is. If Claude can reliably pull last year's budget, draft the plan and post the recap mid-call, the post-meeting workflow shrinks. If consent friction or guest exclusions bite, adoption stalls at internal-only meetings.

The February 10, 2027 free-artifact deadline gives Anthropic roughly fifteen months of beta runway to find out. What happens after that date — and which video platforms follow Google Meet — will determine whether Claude becomes a fixture of the workday calendar or another beta tab closed by policy.