signaldigital.comEst. 1994Berlin — Copenhagen — London — ParisConcept + Design + Technology
, , ,

A Whiteboard That Keeps Working After the Meeting Ends

From the editors

Berlin

Claude Meetings is rolling out in beta for Team and Enterprise plans. It does something genuinely useful: it turns the conversation you are already having into a plan with owners and dates before anyone leaves the call. But it only works with Google Meet, so EU small firms that run on Microsoft 365 cannot use it yet. And the questions about meeting data, retention and workplace rules are still unanswered.

FROM THE EDITORS

Our view is simple. The most valuable part of Claude Meetings is not the AI. It is the meeting discipline it forces. A small team that knows a plan will exist at the end of the call behaves differently during the call. People assign owners and agree dates. They stop saying “we’ll pick this up later”, because later rarely comes. That is worth more than any transcript. But the Google Meet limit is a real barrier for European small firms, and buyers should not ignore the silence on data residency and retention.

The whiteboard that keeps working. Think of the whiteboard in your office. Someone writes on it during a meeting. The meeting ends, the board stays there covered in half-finished ideas, and by Thursday nobody remembers what the arrows meant.

Claude Meetings is an attempt to build a whiteboard that keeps working after everyone has left the room. When the meeting ends, the plan, the timeline and the to-do list already exist. They are not a recording or a transcript nobody reads, but a working document your team can open the next morning and carry on with.

That is the promise. Whether it holds depends on how your firm works, which tools you already pay for, and how comfortable your team is with an AI participant in the room.

The Google Meet problem, stated plainly. At launch, Claude Meetings works with Google Meet only. Anthropic’s own Help Center says Microsoft Teams and Zoom are not supported, and no date has been published for either.

For many EU small businesses, that settles the question for now. If your firm runs on Microsoft 365, as plenty do, from accountancies in Dublin to design studios in Lisbon, this is not a tool you can switch on. For those firms, the nearer route to Claude for now is inside Microsoft 365 Copilot, which we covered in Microsoft Moves Claude Inside the EU Garage.

There is also a difference in the reporting worth knowing. TechCrunch describes Meetings as launching in beta. Unite.AI calls Meetings as a whole “coming soon”, with only the whiteboard part in beta. Anthropic’s own pages are split too: the Help Center refers to Claude Meetings as a beta, while one of its launch articles still says “coming soon”. The fair reading is that Meetings is in beta, with an uneven rollout. Either way, it is not a finished product.

What it actually does during a call. Claude joins as a visible participant and posts a notice in the meeting chat. During the call it can draft the plan being discussed, map a timeline on a shared whiteboard, answer questions in the chat, and pull information from connected work apps. After the call, the notes, action items and anything Claude created are available in claude.ai.

For a small team, the whiteboard matters most. Claude can turn a messy spoken plan into a board with steps, a timeline and dependencies, and when the meeting is over, the board can become a document or a slide deck.

Here is a concrete example. A five-person firm is planning a trade-fair stand. The founder talks through the budget, two colleagues argue about the stand design, and a third mentions that the electrical form has to be filed six weeks before the event. Normally that form is forgotten until three weeks before, and someone pays for an express submission.

Here, the conversation produces a timeline on the board instead: design sign-off by one date, the electrical form by another, shipping booked by a third, each with a name next to it. The plan exists before the coffee cups are cleared. The value is not the AI as such. It is that the plan becomes explicit while the people who made the decisions are still in the room.

The external-guest rule is a sensible default. By default, Claude does not join calls with external guests, and it leaves automatically if one joins. An admin can change this.

For a small firm, this is the right default. Client calls and supplier negotiations often involve commercial information that should not go to a third-party AI without a deliberate decision. If you want Claude in a client call, someone has to choose that on purpose. That is how it should be.

The questions the sources do not answer. This is where EU small firms need to slow down.

An AI participant that listens to a meeting raises questions about legal basis, transparency to participants, and where meeting data is stored and for how long. The published material says nothing about retention periods, storage location or EU data residency for Claude Meetings. That is not a minor omission. For a firm handling client information, HR discussions or sensitive negotiations, these are the first questions to ask before switching anything on.

Some EU countries also have rules on monitoring employees, and firms with works councils or staff representatives may need to involve them before an AI tool listens to internal meetings. This is not legal advice, and the details vary by country and workplace agreement. The general point is simple: an AI participant changes how meetings are observed and recorded, and in some workplaces that change needs a conversation with staff before it needs a credit card.

Counterarguments worth taking seriously. A meeting that produces a plan automatically can also produce a confident plan nobody actually agreed to. Claude drafts what it heard. If the discussion was messy, vague or full of unresolved disagreement, the result may look tidy and still be wrong. Someone still has to check it. The tool does not remove the need for judgement. It removes the excuse that nobody wrote anything down.

Pulling figures from connected apps into a call also means more data in more places. Anthropic says Claude works within your existing permissions and that admins can limit what each connector does. That describes the design, not your own set-up, so check the permissions yourself.

People may speak less freely with an AI participant in the room. A team that is candid about problems in private may become more guarded. Small firms survive on candour, so if the tool makes people quieter, it is not working.

Then there is lock-in. Plans, whiteboards and sheets that live inside one AI vendor’s workspace are harder to leave behind. Anthropic says boards can be turned into documents or decks, exported as images, or sent to Lucid, Miro and Zoho Vani, which helps. But the working surface itself is Claude’s, a trade-off we also raised in our piece on Claude Fable 5.2. Make that choice consciously.

Finally, there is no price yet. Meetings is for Team and Enterprise plans only, so solo founders on individual plans cannot use it, and no prices have been published for Meetings itself. A small firm already paying for Google Workspace or Microsoft 365 and a project tool should ask what this adds that its existing tools do not.

Back to the whiteboard. A whiteboard that keeps working is a good idea. Whether it works for your firm depends less on Anthropic’s roadmap than on where your meetings happen, who is in them, and what you are prepared to ask before you switch it on.

What to do this month

  • Check your video set-up. If your firm runs on Microsoft Teams, Claude Meetings is not available to you, so do not plan around it. If you use Google Meet on a Team or Enterprise plan, ask about beta access and pricing.
  • Ask three questions before enabling it: how long is meeting data kept, where is it stored, and is there EU data residency? If the answers are unclear, keep client calls out of it.
  • If you have a works council or staff representatives, talk to them first. An AI participant in internal meetings is a workplace change, and in some EU countries that conversation is not optional.
  • Run one internal meeting with it and compare the output with what was actually agreed. If they do not match, you have learned something about the tool, and about how your team talks.
  • Keep the external-guest default as it is. Client and supplier calls should need a deliberate decision before an AI participant joins.

Sources

— FROM THE EDITORSSignaldigital

Read next from the editors

Leave a Reply