Before you put your data into any AI assistant

Ask one question:
who else is going to read this?

A contract under negotiation. A payroll sheet. A quote for a competitor. A letter from your lawyer.
That isn't “data.” That's your business.
And every free assistant you used today wasn't free — you paid in a different currency.

We sell you software. We don't sell your data.

That's why we charge. Privacy costs — we chose to be the ones paying.

Our commitments

Five commitments, in plain words.

Four locks. Not one switch.

Turn on Confidential mode for a sensitive project, and four gates close at once.

  • No memory saved
  • No file added to your Documents
  • No chat attachment saved as a document
  • No task or file created from the chat

One setting is easy to forget. Four locks in four places are not.

Off until you turn it on.

Error reports are disabled by default. We collect nothing until you decide.

And if you enable them: the screenshot is of the WALAA AI window alone — not your desktop, not your other windows, not your passwords, not your keys.

Your mail stays yours.

Connected by your permission. Revoked in one click. We don't sell your addresses, share them, or use them for anything you didn't connect them for.

Isolation in the database, not on the screen.

Your account is isolated row by row. No account sees another's data — because the database refuses, not because the interface hides it.

The difference between the two is the difference between a lock and a curtain.

Simple counting questions in Arabic are answered on your machine.

“How many tasks are overdue?” · “How many unread emails do I have?” · “How many meetings do I have?” — these counting questions are answered on your machine without AI, and aren't sent to the cloud, because they don't need it. This works when you ask them in Arabic. Every other question is answered in the cloud.

Ask any assistant these questions before you trust it:

Can I stop it from keeping one project's files?
Yes — Confidential mode on that project: nothing from it is kept in Documents or memory, but the answer is still written in the cloud
Does it send reports about me without my knowing?
No — off by default
If it captures my screen — what exactly does it capture?
Its own window only
Does it read my mail by my permission or its company's?
By yours — revoked whenever you want
Can another account see my data?
No — isolation in the database
Does it send every question to the cloud?
No — ask in Arabic how many tasks, unread emails or meetings you have, and the answer comes from your machine

Keep this list. Ask it of the others.

And with rare honesty for this market:

Text generation happens in the cloud — as in most AI assistants today, and if someone tells you otherwise, check their claim. What's sent is only what the current request needs: your question, the file or message you asked about, and whatever in your records bears on it — never your whole archive, never your whole mailbox.

And before anything leaves, the app strips identity from it on your device: the names of contacts and projects it knows, and what it recognises as a name after a title like “Eng.” or a word like “Company”, and emails, phone numbers, ID, account and card numbers — each replaced with a token that points to no one, and put back into the answer when it reaches you. So the cloud sees that a person signed with a company — not who they are. Numbers and amounts stay, because the arithmetic needs them. Voice in dictation and voice chat is not masked, and neither are scanned page images when AI reads them.

We won't promise what we don't do. We promise what you can test yourself.

Go to Support