Most AI meeting tools now call themselves private.

Usually that means your data is encrypted on the way to a server. Sometimes it also means it is encrypted while stored there.

That is useful. But your meeting still left your device.

And with meetings, that matters.

Meetings are not normal documents. They contain unfinished thoughts, client context, internal politics, legal questions, hiring decisions and numbers that are not ready to be shared. A transcript captures all of it.

So the question is not only whether the data is encrypted.

The better question is: why does this meeting need to leave the laptop at all?

Security agencies make a similar point more broadly. When sensitive data is sent to a third party, organisations need to understand how that data is handled and whether it fits their regulatory duties. [1]

Privacy policies should not become your job

As a founder, consultant or advisor, you do not have time to dig through every line of every cloud-based privacy policy before a sensitive call.

What happens to the audio? Where is the transcript stored? Can anonymised data be used for model improvement? Which third parties process it? Who can access it?

Even when the policy is reasonable, the work is pushed onto the user.

Recent coverage around AI note-taking tools shows how easily “private” can mean something different from what users expect. In one case, notes described as private could still be shared by link, while anonymised data could be used for model improvement unless users opted out. [2]

Bots also change the room

There is a human side too.

A meeting bot is not invisible. It enters the call. It asks for permission. Sometimes it arrives before the person who invited it.

That is awkward. I have experienced it myself.

For confidential conversations, it also changes the feeling in the room. People may speak differently when a visible recorder is present.

Legal commentators have also warned that AI note takers can create issues around consent, confidentiality and privilege in sensitive meetings. [3]

What Weeve does differently

Weeve is built around a simpler idea.

Your meeting content should stay on your laptop.

The recording runs locally. The transcription runs locally. The speaker detection runs locally. The summary runs locally.

Your audio is not sent to us. Your transcript is not sent to us. Your summary is not sent to us. Your decisions, tasks, templates and meeting titles are not sent to us.

Not encrypted and uploaded. Not redacted and uploaded. Not anonymised and uploaded.

It stays on your device.

Why we removed cloud summarisation

For a while, Weeve had an optional cloud path.

Users could connect their own Anthropic, Google or OpenAI account and let one of those services handle summarisation. It was off by default. Almost nobody used it.

That told us something.

People were choosing Weeve because they wanted the cloud out of the loop.

So we removed the option.

Now the answer is simple: Weeve does not send your meeting content to cloud AI providers.

Offline is part of the point

Once Weeve has downloaded the models it needs, it works without an internet connection.

On a plane. On a train. In a client environment with strict IT rules. In a place where the Wi-Fi is simply bad.

Your notes still get created because the core product runs on the machine in front of you.

For many teams, that is the difference between a nice tool and an allowed tool.

Sources

[1] UK National Cyber Security Centre, “Dealing with data.”  

[2] The Verge, “PSA: Anyone with a link can view your Granola notes by default.”  

[3] Smith Anderson, “The Silent Guest in Your Meetings: Legal Risks of AI ‘Note-Takers’.”