Claude Meetings enters beta and puts Anthropic in competition with Otter, Fireflies and Granola

Anthropic published its announcement of Claude Meetings on November 12, 2026, a beta feature that puts Claude as a visible participant inside video calls. According to material released by the company, the assistant joins the meeting like any other guest, follows the conversation in real time, and starts producing work (a document, meeting minutes, an action spreadsheet) while people are still talking.

The launch arrives in the same package of November 10 and 12 announcements that included self-hosted Claude Managed Agents and Claude for Manufacturing, a move that shows Anthropic pushing vertical products on top of its own API. This is the context that matters for anyone who has already built workflows on top of Claude: the company is now competing directly with product layers that, until now, were the domain of third parties or homegrown integrations.

How it works in practice

The example described by Anthropic itself involves a team planning a customer event. Someone asks out loud for Claude to "start a document for this plan"; the assistant writes down what has already been discussed, and the other participants edit the text live while the conversation continues. When the topic turns to budget, someone asks how much last year's event cost, Claude asks for permission to check connected tools, posts the figure in the call's chat, and compares it with the current plan.

Further on, when the topic is scheduling, someone asks to put everything on a whiteboard, and the team starts working on it in parallel with the document. Before wrapping up, someone asks for a recap, and Claude summarizes what was decided. After the call, all the material (minutes, action items, document, whiteboard) remains available on a dedicated meeting page inside claude.ai, accessible afterward for questions about what was discussed.

This is the central point for anyone who codes: the command interface is speech itself or the call's chat, not an API you control. In short: Claude Meetings is a product, not an SDK, and that changes the kind of decision technical teams need to make.

What this changes for those who built on top of the API

Teams that have already integrated Claude via API to summarize meetings, generate minutes, or automate follow-ups (combining, for example, a transcription tool with API calls for summarization) now face a direct question: keep their own integration or migrate to the native feature?

The advantages of the native path are clear:

• No transcription and orchestration pipeline to maintain;
• Claude appears as a visible participant, which reduces approval friction among teams already wary of silent bots recording calls;
• During the beta, according to Anthropic, notes, transcripts, and in-call responses are free on Team and Enterprise plans, except when a response uses a connected tool, in which case it counts against the plan's usage;
• Documents, whiteboards, and other artifacts created during calls remain free until February 10, 2027, according to the official page.

The downsides, meanwhile, weigh heavily for those who need fine-grained control: Claude Meetings currently only works with Google Meet (other video conferencing platforms are "on the way," with no date given), and the decision flow (who approves the use of a tool, what becomes meeting minutes, how the artifact is versioned) is defined by Anthropic, not by the team that integrates it.

Security and the gaps that still exist

For security architects, the most relevant detail in the announcement is the list of what the beta still doesn't cover. Anthropic states that Claude Meetings is not available for organizations with zero data retention (ZDR), HIPAA-ready configuration, or customer-managed encryption keys (CMEK). That takes the feature off the immediate radar of healthcare companies, regulated finance, and any stack that depends on these guarantees, precisely the customer profile most likely to have already invested in building its own meeting automation via API, with full control over data retention.

The privacy model described works like this: each meeting gets its own page on claude.ai with notes, transcript, and any artifact produced, visible and editable by anyone in the organization who was invited. By default, Claude stays out of calls with external guests and leaves if someone from outside joins mid-meeting; an admin can enable this. When enabled, Claude posts a notice in the call's chat but doesn't respond there, although whatever external guests say does go into the notes and the transcript.

Another technical point: Claude saves transcript, summary, and screen descriptions, not audio. That reduces the surface of sensitive stored data, but it also means any later audit depends on the quality of the transcript, not the original recording.

The partners that are also competitors

One of the most revealing passages in the announcement is the list of partners cited by Anthropic: Gong, Attention, Granola, Fireflies.ai, Otter, and Plaud. The company says it is working to soon allow sending a meeting's transcript to these tools that teams already use.

In practice, Anthropic is building a feature that overlaps with what these same companies sell as their main product (automatic minutes, meeting summaries, transcript search), while at the same time announcing integration with them. It's a common stance among platforms trying to become an infrastructure layer without alienating the ecosystem that runs on top of them, but it doesn't change the fact that, for the end user, the decision of which meeting tool to use has become more contested.

Who decides access, and how much it costs after the beta

On Team plans, anyone can invite Claude to their meetings by default, and an admin can disable that for the organization. On Enterprise plans, it's the opposite: an admin needs to turn on the feature, fully or partially, for the team. This difference in default posture (opt-out on Team, opt-in on Enterprise) is a sign that Anthropic treats the feature as more sensitive the larger the organization.

What the source doesn't answer is the price after the beta. The free status of notes, transcripts, and artifacts has a deadline (February 10, 2027 for the artifacts) or a condition (using connected tools already counts against the plan today), but there is no published post-beta pricing table. For teams weighing whether it's worth migrating an already-stable automation to Claude Meetings, this is the gap that weighs most heavily: getting in now, for free, is easy; predicting the recurring cost of using this at scale is not yet possible.

What still remains to be covered

Anthropic has not detailed latency, context limits for long meetings, or how Claude handles multiple people talking at once or switching topics quickly, a common situation in real technical meetings. There is also, for now, no mention of support for Zoom or Microsoft Teams, the two platforms most used in Brazilian corporate environments alongside Google Meet. Until these points become clear, Claude Meetings is more a signal of Anthropic's strategic direction than a safe replacement for anyone who already has meeting automation running in production.