makeprivacy.com · private business systemsLooking for free guides and browser tools? makeprivacy.org ↗
makeprivacy.comContact

Everyday problems · Anyone holding client data

Can you use AI on client data without handing it over?

Yes, if the AI works inside accounts you control, under business terms that bar training on your data, and only sees what it needs for each question.

What usually happens

The common setup, and its catch

  • AI built into the email and storage suite. Convenient because it reads everything, which is exactly the trade.
  • Staff pasting into consumer AI apps. Client data leaves the company one copy-paste at a time, under terms nobody checked.
  • A chatbot connected to every account. Broad access, and actions taken without anyone seeing them first.

What we set up

In accounts your company owns

  • An assistant that answers from your own records, with the permissions of the person asking
  • Search indexes and voice transcription running in your own cloud account
  • AI only through providers whose business terms bar training on your data
  • Every action prepared for a person to confirm, on a page the AI can't press

You can try the approach on this site: the assistant on the home page runs on an AI model in our own cloud account, answers only from this website, and doesn't store your questions.

Already running

Live in our own companies

See our work.

An assistant that asks first

A private AI assistant in our own account. It prepares the work, and a person confirms it.

Every client conversation in one place

45,610 messages from email and chat apps, in an archive we own, kept even when someone deletes them.

Out of Google in 36 hours

Our company mail, files and the automations around them moved out of Google Workspace, with nothing lost.

A first step

Where we'd start

Usually search across your own records first, then actions, each one confirmed by a person. Every project is quoted after a conversation about what you run today.