Swiss, privacy-first AI chat
The leading AI models. Without a record of who asked.
ZeroLog.ai is a privacy layer between you and the providers. Your conversations are stored on your own device — there is no permanent server-side copy of what you typed or what came back.
The privacy benefit is not that the provider cannot read the request. It is that the request is separated from your direct consumer identity.
Stored on your own device
Your conversations are saved locally, in your browser. The server never receives a copy to keep.
How much, never what
Your account is linked to the number of tokens you spent. It is never linked to the words you spent them on.
A request from ZeroLog.ai
The provider sees a request from ZeroLog.ai under one shared service identity — not a person at an address.
A privacy layer between you and the provider.
The provider still needs the content of your request to answer it, and it does so without managing your individual consumer account. That separation is the product, and it is enforced in the code rather than promised in a policy.
What we hold
- Your email address and the password you set.
- An active sign-in session.
- A usage counter: tokens spent, requests made.
- A 90-day log of sign-ins, sign-outs and password resets.
What we never hold
- What you asked, in any form.
- What the model answered.
- Any record linking your account to either — not even briefly.
- Your files. A PDF you attach is read on your device and stays there.
How ZeroLog.ai works
Sign in with an email and a password.
That is the whole account. No name, no phone number, no address book.
Type your message.
The conversation is written to your own browser storage as you go.
Your turn is sent through our Swiss servers.
The conversation travels with it, because the server keeps nothing of it between turns.
We pass it to the provider under one shared service identity.
Details that identify you are stripped before the request leaves, and there is no other route out.
The answer streams back to you.
When the stream closes, the request is gone from the server. Your usage counter goes up by what it cost.
All the leading models. One private interface.
One account, one interface, and a model picker on every conversation. Choose the model that suits the question, and switch when it stops suiting it.
In the picker today
- Claude Sonnet 5
- Claude Opus 4.8
- Claude Haiku 4.5
- GPT-5.5
- DeepSeek V4 Flash
What comes with it
- Attach a PDF: its text is read on your device, and no bytes are uploaded.
- Export your whole history to a file, in one click, whenever you want it.
- Point ZeroLog.ai at a folder you already control — iCloud Drive, Dropbox — and your history follows you between your own devices. The folder is yours.
Efficiency over excess.
Running a model costs energy. We intend to be straight with you about what ours costs, with figures you can check, rather than decorate this page with a leaf.
Swiss jurisdiction. Swiss legal process.
Your account, your session and your usage counter live in our own Swiss database. Production is planned to run in AWS's Zurich region, so a request leaves from Switzerland and is answered back into it.
Your chats should belong to you, not to the platform.
Clear control over what is retained, and control that stays with you.