Your Meetings Are Not Training Data: The Laxis No-Training Pledge

Your meetings contain the most honest version of your business. Pricing you haven’t announced. A roadmap quarter you haven’t shipped. The customer who is about to churn, the candidate you are about to hire, the term sheet you are about to sign. People speak in meetings the way they never write in documents — because they believe the room is the room.

So here is the Laxis position, in plain words, with no asterisks:

The pledge

Laxis will never use your data to train AI models without your consent.

Not your meeting recordings. Not your transcripts. Not your notes, summaries, or action items. Not the words you dictate with our voice keyboard. Training on customer content is never on by default, never buried in a settings page as something you must discover and switch off, and never a condition of using Laxis. If we ever ask to use content for training, it will be exactly that — an ask: a clear, specific, opt-in request that you can decline or withdraw from at any time, where saying no changes nothing about your service. Until you say yes, the answer is no.

Notice what this pledge is about: what we do with your data. That is the promise we can make without qualification, so that is the promise we make.

This is not a marketing claim floating free of our legal documents. It is written into our Privacy Policy (Section 3, “AI Model Training — Our Commitment”) and repeated in our Terms of Use (Section 8), where we commit that your content is licensed to us only to operate the service for you. If we ever proposed to change this commitment, our policy requires us to tell you before the change takes effect — and it would never apply retroactively to content you created before it. We do not expect to ever make that proposal.

Why we are drawing this line

AI notetakers are genuinely useful. That’s why the training question matters so much: the more of your working life flows through a tool, the more valuable — and the more sensitive — its data exhaust becomes.

The industry norm, quietly, has been to treat that exhaust as an asset. Read the fine print at major vendors and you will find some combination of: training on by default, opt-outs buried in settings, opt-outs that only work org-wide on the most expensive tier, and acknowledgments that once your words are inside a model, they cannot be taken back out. The consequences are no longer theoretical — as we documented in AI Notetakers in Court, a federal judge recently let a class action proceed against Otter.ai on the theory that a vendor which “independently collects, retains, and uses communications for its own commercial purposes” is not a neutral tool but a third-party eavesdropper. Using conversations to improve the vendor’s own models is among the uses alleged. Granola is facing a class action of its own, alleging silent capture and training by default.

We do not want our business anywhere near that line — not legally, and not morally. A meeting tool is a guest in the most confidential conversations a company has. Guests do not take notes for themselves.

What we do instead

To be precise about the boundary:

• To run the service for you, we process your content: transcribing your meeting, writing your summary, syncing your action items to the CRM you connected. That processing happens on your instructions, for your benefit, and the output belongs to you.
• To improve Laxis as a product, we use only de-identified or aggregated data — the kind that tells us “the summary feature failed 2% of the time on Tuesday,” never what anyone said.
• We do not sell your personal information.
• Your notes are portable. Export summaries and download transcripts whenever you like. And deletable: delete items in the app whenever you like; delete your account and your content is deleted immediately, backups included.

Don’t take our word for it — check

A trust claim you cannot verify is just copywriting. So:

• Read the commitment where it legally lives: Privacy Policy §3 and Terms of Use §8.
• See who touches your data: our subprocessor list is available on request at [email protected].
• Business customers can request our Data Processing Addendum and review our SOC 2 posture before signing anything.
• Hold us to it publicly. This post is dated, signed by the company, and written to be quoted back at us.

The bottom line

There are two ways to build an AI company. One treats every conversation it touches as raw material. The other treats every conversation as a confidence. Laxis is the second kind of company — and we have put it in writing, in the documents that bind us, so that it does not depend on who happens to be running the company blog this quarter.

Your meetings are yours. We just take the notes.

Read next: AI Notetakers in Court: What the Granola and Otter Lawsuits Mean for Your Meetings · Is It Illegal to Record a Conversation? · Laxis Privacy Policy