Claude Meetings is now available: Add Claude to your video call and get

Anthropic launched Claude Meetings in beta, enabling Claude to join video calls as a visible participant that tracks live conversations, fields team questions, and begins project work directly inside meeting sessions. According to Claude Blog's report, the feature is rolling out to organizations on Team and Enterprise plans, initially supporting Google Meet with additional video conferencing integrations planned for future releases. During active calls, participants can query Claude aloud or through written chat, prompt it to draft collaborative documents or whiteboards in real time, and rely on the assistant to log transcripts, summaries, and action items automatically.

This report examines the operational architecture, security controls, partner integrations, and rollout requirements for engineering teams and enterprise administrators deploying Claude Meetings across their organizations. Anthropic designed the assistant to answer written queries during meetings while pulling live records through connected organizational tools whenever participants grant explicit approval. The system provides teams with post-call workspaces in claude.ai, safeguards cross-organization call boundaries by default, and maintains strict administrative exclusions for accounts operating under specific regulatory compliance profiles.

01What shipped in Claude Meetings

Claude Meetings introduces an interactive AI agent that joins video sessions alongside regular attendees, announces its presence in the call chat, and responds to spoken or text prompts from participants. During collaborative planning sessions, attendees can instruct Claude to initialize project documents or construct timeline whiteboards while conversation continues. Teammates listed on the calendar invite can view, revise, and co-edit these artifacts simultaneously without leaving the meeting flow.

The assistant references external company information during calls by accessing authorized workplace tools after requesting explicit user confirmation. For example, when attendees ask Claude to retrieve historical budget figures, Claude prompts the speaker for permission, checks connected services, and outputs the comparative findings directly into the chat stream. Claude also generates comprehensive recaps, summaries, action checklists, and full meeting transcripts that automatically populate dedicated meeting records in claude.ai once calls conclude.

Attendees can review past calls and query Claude about historical decisions directly from their claude.ai accounts or schedule Claude for upcoming sessions straight from the integrated calendar interface. Anthropic is also building partner workflows with external meeting and conversation platforms, including Gong, Attention, Granola, Fireflies.ai, Otter, and Plaud. These integrations will enable organizations to route meeting transcripts generated by Claude into their existing third-party analysis and note-taking stacks.

02What improved in Claude Meetings

Anthropic changed how meeting collaboration tools manage audio, participant access, and external data queries across active enterprise video calls. Instead of recording and retaining raw audio streams, Claude Meetings saves text transcripts, meeting summaries, and screen descriptions to preserve participant privacy while delivering structured outputs. The system also introduces individual tool approval gates, requiring participants to confirm permissions before Claude retrieves data on their behalf during live discussions.

Under the beta rollout, meeting transcripts, automated summaries, and text answers generated by Claude incur no additional charge on Team and Enterprise plans. Answers that query external connected tools consume standard plan usage units based on existing organization allowances. Furthermore, collaborative documents, whiteboards, and associated artifacts produced during calls remain free of charge across all eligible plans through February 10, 2027.

03What you gain from Claude Meetings

Organizations gain automated post-call documentation coupled with strict internal boundary controls that protect proprietary conversations from unintended disclosure. Claude generates a distinct workspace page in claude.ai for every finished meeting, housing the full transcript, notes, and interactive artifacts created during the session. Internal team members who were on the meeting invite receive immediate viewing and editing rights, with the option to share materials across their wider organization.

Security policies prevent data leakage to external parties by defaulting to an internal-only posture. Claude automatically declines to enter calls that include participants outside the host organization, and it disconnects immediately if an external attendee joins the session while it is in progress. Enterprise administrators can choose to override this restriction, allowing Claude on calls with external guests while silencing Claude's chat replies and preserving external guest remarks in the meeting transcript.

Permission isolation ensures that Claude only queries connected data sources with explicit approval from the specific individual making the request. Claude asks for permission upon the first lookup attempt, applying that consent to subsequent automated operations for that user. When participants issue voice commands aloud, Claude verifies the identity of the speaker before proceeding with data queries, ensuring that answers posted in the shared chat remain accountable to the proper team member.

04How to get Claude Meetings

Claude Meetings is accessible in beta on Google Meet for organizations holding active Team and Enterprise subscriptions, while setups with customer-managed encryption keys, HIPAA configurations, or zero data retention policies remain excluded. On Team tiers, meeting invitations are active by default and can be disabled through administrative controls, whereas Enterprise administrators must explicitly enable the feature across organizational units. Developers using the command line can update and verify their local Anthropic tooling environment using standard terminal commands:

npm install -g @anthropic-ai/claude-code

After installing or updating the client suite, developers can check active installations, authenticate, or adjust workspace models. To switch models inside an active terminal session, use the interactive model selection prompt:

/model claude-fable-5-1

Teams launching CLI sessions with explicit model pins or checking administrative configurations can execute direct CLI commands to establish targeted defaults:

claude --model fable

05What to watch after Claude Meetings

System administrators and team leads must assess organizational readiness, participant consent obligations, and compliance requirements before enabling automated call attendance. Anthropic requires organizations to ensure that all internal and external attendees approve of automated transcription before deploying Claude into live calls or enabling automated calendar joining. Organizations subject to strict data governance frameworks must confirm whether their specific retention and encryption architectures permit beta participation.

Enterprise teams must also track the rollout timeline for additional video conferencing ecosystems beyond Google Meet, as multi-platform support remains under active development. IT managers should coordinate with department leads to review connected tool integrations, verifying which data sources team members can authorize during calls. Monitoring tool usage will also allow teams to manage subscription unit consumption as users query integrated databases during active meetings.

Finally, organizations utilizing transcription platforms like Gong, Attention, Granola, Fireflies.ai, Otter, and Plaud should prepare for direct transcript synchronization pipelines as Anthropic finalizes partner connectors. Integrating these pipelines will allow teams to feed structured notes, decisions, and action items directly into CRM and customer intelligence workflows. Administrators can review detailed technical configuration guides, permission hierarchies, and provisioning procedures inside the official Help Center documentation.