Anthropic launches Claude Meetings: AI joins Google Meet and starts working on tasks right away

A familiar picture: an hour-long call, everyone nods, and afterward someone has to remember what was agreed, rewrite it as tasks and draw up a timeline. Usually that work starts once everyone has already turned their cameras off. Anthropic proposes moving it into the meeting itself: this Tuesday the company introduced Claude Meetings and several other office tools.

What Claude Meetings can do

Claude joins a video call as a visible participant. It listens to the conversation and does work in parallel. Say a team is discussing a campaign launch. While people argue over deadlines, Claude sketches a plan, lays it out on a whiteboard by date and answers questions in the meeting chat. If someone asks how much was spent last year, it goes into the connected work apps and fetches the figure.

After the call, the notes, the list of action items and everything Claude created remain in claude.ai. What you get is not just minutes but a set of drafts you can keep working with.

The feature has built-in safeguards. Claude introduces itself as a participant and says so in the chat. It does not join calls with external guests without administrator permission, and if a guest appears in the middle of a meeting, it leaves automatically. For companies with client calls, this is an important detail.

For now this is a beta for the Team and Enterprise plans, and only Google Meet is supported. About other platforms the company says only that they will come later.

Whiteboards, spreadsheets and data around it

Meetings is only the tip of the set. Three more things were released with it.

Claude Whiteboards is a shared board where the team discusses a plan and Claude turns it into a structured space with timelines and deadlines. Anthropic fairly notes that many people already brainstorm with Claude, but to share the result they have to copy it somewhere else. Here the team works on one board in real time, and the finished board can be turned into a document or a presentation. Inside Meetings, the board grows right as the conversation unfolds.

Claude Sheets is collaborative spreadsheets. Ask it to build a campaign tracker or a hiring budget, and Claude creates a working spreadsheet: several tabs, formulas, formatting, dropdown lists. The team then edits it together with Claude. Claude edits cells on equal footing with humans and describes every change, so you can see exactly what it changed. For spreadsheets where a formula error costs money, this is the most useful part.

Both tools are available in beta on paid plans on the web, on desktop and on mobile devices.

Claude Data (formerly Claude Dashboards) has exited beta. It connects to wherever a company stores its numbers, such as a data warehouse or a CRM. You can then ask questions and get back dashboards, text "data stories" or videos with voiceover. In addition, interactive connectors have appeared: full versions of apps like Adobe, Atlassian and HubSpot open right next to the chat, and you can ask Claude to work inside them.

Why it matters and what to do about it

The main signal is not in the individual features but in the direction. Anthropic, OpenAI and Meta (with its Muse agent, which builds mini-apps for a specific task) are competing for the role of a new kind of office suite, where AI works in the same place as the team rather than answering in a side window. This is a direct challenge to Microsoft and Google.

For product managers and founders, the practical takeaways are:

• If your team lives in Google Meet and is on Team or Enterprise, a pilot is worth trying on internal planning meetings, where the risk is minimal.
• Agree on rules with your admin in advance: who is allowed to invite Claude and how to handle meetings that include external people.
• Check the numbers Claude pulls from connected systems. A beta is a beta, and a figure in a campaign plan lives a long time.
• If you are building your own collaboration product, keep in mind that "copy from the chat to somewhere else" is already considered a flaw, and competitors are closing that gap.

The limitations are also obvious. Only Google Meet, only beta, and there is no public data yet on how it performs in live conversation with interruptions and arguments. Whether Claude becomes a useful partner in a meeting or just another participant who gets in the way will only become clear from real teams. But the idea that "work starts during the discussion, not after it" looks clear enough to test on a single call.