Back to News

Managed IT · Nov 13, 2026

AI assistants are joining your meetings. Set the ground rules before they do

Most meetings end the same way. Someone says "I'll send notes," the notes arrive late or not at all, and half the action items quietly disappear. It is a familiar pattern on small teams.

That is exactly the problem the big AI companies are now going after, and they are doing it by putting the assistant directly in the call.

What changed this week

On November 10, Anthropic announced Claude Meetings, now rolling out in beta on its Team and Enterprise plans. According to Anthropic's announcement and help center, Claude can join a Google Meet call as a visible participant. It follows the conversation, writes the notes and action items, answers questions out loud or in the meeting chat, and can build documents, whiteboards, and other materials with the team while the discussion is still going. After the call, everything it made lives on a meeting page people can go back to and ask questions about.

A few details stand out:

  • It can use connected tools. Claude can use a person's connected work apps during the call, such as sending a Slack message, with that person's permission. Anthropic says it asks the first time, and a yes applies going forward.
  • It can auto-join. People can set Claude to join all their meetings, or all except one-on-ones.
  • It is on by default for some plans. On Team plans, inviting Claude to meetings is available by default and an owner can turn it off. On Enterprise plans, an owner turns it on for all or part of the organization.
  • External guests are handled carefully. TechCrunch reports that Claude identifies itself in the meeting chat, will not join calls with external guests unless an admin authorizes it, and leaves automatically if an external guest joins.
  • It is not for every organization yet. During the beta it is not available to organizations using zero data retention, HIPAA-ready configuration, or customer-managed encryption keys.

Anthropic's own help center puts it plainly: before you add Claude or turn on auto-join, make sure everyone on the call is okay with being transcribed.

Anthropic is not alone. Grok Bot 1.0 also lets teams bring AI teammates into their video calls. The direction is clear. AI is moving from a chat window you open to a participant that sits in on your work.

Why this is a Managed IT question, not just a productivity one

A meeting assistant is useful because it hears everything. That is also why it needs rules.

Think about what gets said on an ordinary week of calls. Payroll numbers. A customer's account details. A conversation about an employee. A pricing discussion you would not want forwarded. Once an AI is transcribing those calls and can reach into connected apps, the meeting is no longer just a conversation. It is data, stored somewhere, searchable later, and possibly able to trigger actions in other systems.

None of that makes these tools bad. It means the decisions should be made on purpose, by the business, before the first auto-join instead of after.

Five ground rules to set first

1. Decide who can invite the assistant. If a feature is on by default, the first decision is whether it should stay on. For many teams the right start is a small pilot group, not the whole company.

2. Write a consent norm and stick to it. Tell people when an AI is in the room. Some states require consent from everyone on a recording, so check the rules where you and your callers are and get advice if you are unsure. A simple habit helps: the host announces it at the start, every time.

3. Separate meetings by sensitivity. Internal stand-ups and project check-ins are good candidates. HR conversations, legal matters, and anything involving protected client information usually are not. If your business works under compliance requirements, confirm whether the tool's current setup supports them before anyone turns it on.

4. Limit what it can touch. Connected tools are where the time savings live, and also where mistakes travel fastest. Start with read-only access where you can, keep approvals on for anything that sends messages or changes records, and review which accounts are connected.

5. Know where the notes go. Transcripts and meeting pages are records. Decide who can see them, how long they are kept, and who owns them when someone leaves the company.

Get the action items out of the meeting

Notes are only half the value. The other half is what happens next. A meeting summary that sits on a page nobody checks is just a better-formatted version of the old problem.

The teams that get the most from this will connect meeting output to the systems they already run on: tasks into the project board, follow-ups into the CRM, decisions into the shared drive where people actually look. That is workflow design, and it matters more than which assistant you pick.

Where Centervert fits

This is the kind of work we help businesses with. On the Managed IT side, we review admin settings, connected accounts, and data retention before an AI tool goes live, and we help write the plain-language policy your team will follow. On the automation side, we help teams route meeting notes and action items into the tools they already use, with tools like n8n, Zapier and Composio.

The one thing to do this week

Check whether any AI meeting assistant is already turned on in your accounts, including ones employees may have added on their own. Then pick one recurring internal meeting, set the ground rules above, and run a pilot for a month. Think First, AI Second. If you would like help setting it up, we are glad to take a look.

Sources