Anthropic has moved its Claude assistant out of the chat window and into live video meetings. The company's Claude Meetings beta lets Claude join a Google Meet call as a visible participant, where it can answer questions, draft plans and map out timelines while people talk. Anthropic unveiled the feature on November 10, 2026, and it marks one of the clearest attempts yet to put an AI agent inside the daily work of teams rather than beside it.

From Note-Taker to Participant

Most meeting tools have stayed in the background, transcribing conversations and summarizing them afterward. Claude Meetings is built to do more during the call itself. Coverage of the launch describes Claude following the discussion in real time and starting on tasks as people talk, including drafting plans under discussion, answering questions posted in the meeting chat and building timelines as they take shape.

The shift is subtle but significant. A transcript records what was said. An agent in the room can change what happens next, such as flagging an action item that has no owner or drafting a timeline before the meeting ends.

The feature can also draw on information outside the call. Claude can pull data from connected work applications, such as customer relationship management systems or data warehouses, but only when a user has granted that permission. By default it responds in text rather than speaking.

What Teams Get Afterward

Once the meeting ends, notes, transcripts, action items and any documents Claude produced are gathered on a dedicated page in claude.ai. That turns the output of a call into something a team can act on the next morning, rather than a recording nobody revisits.

Anthropic is also beta testing Claude Whiteboards and Claude Sheets, which suggests the company wants meeting output to flow into the same workspace as other documents and spreadsheets.

Access, Pricing and Guardrails

The beta is limited to Team and Enterprise subscribers. One report puts Team pricing at $25 per user per month and sets a $400 monthly minimum for Enterprise, though buyers should confirm current terms on Anthropic's pricing page. Google Meet is the only supported platform at launch, and Anthropic has said other video services will follow.

The guardrails matter most to administrators. Claude will not join a call with external participants unless an admin approves it, and if an unapproved outside guest joins mid-call, Claude leaves on its own. Organizations running HIPAA or customer-managed encryption configurations cannot use Meetings yet.

Why It Matters

The launch shows where agent products are heading. Agents are moving from answering single prompts toward acting inside workflows that unfold over time, and a meeting is one of the messiest places for that to happen. An agent in a call has to track several speakers, decide what counts as an instruction and avoid sharing information with people who should not see it.

That is why the admin controls are as important as the features. Companies adopting these tools will need clear rules about who can add an assistant to a call, what it may access and how its notes are stored. For teams already using AI for documents and research, Claude Meetings extends that workflow into the live conversation, where many decisions actually get made.

The timing also matters. Emergent reported that the launch coincided with the start of Anthropic's IPO roadshow, so the feature is arriving as the company is under heavy public scrutiny.

What to Watch

Two questions will determine how far the feature goes. The first is platform coverage. Anthropic has named only Google Meet so far, so organizations on other meeting systems will have to wait. The second is autonomy. At launch Claude responds to what it is asked and uses connected tools only with permission. That limits risk but also limits value. Each new connector widens what an agent can see during a call, so the permission model will be worth watching as the beta matures.

Sources