Frequently asked questions
What we hold, what the provider sees, and where the boundaries of this are. If an answer here reads as a hedge, that is deliberate — we would rather be exact than reassuring.
What information could you provide about me if someone asked?
Very little. Your email address, when you signed in, and how many tokens you have spent.
We hold no record of what you asked or what the model answered, so there is nothing of that kind to hand over. We can only provide what exists.
Does the provider see my messages?
Yes. The provider needs the content of your request to generate an answer.
What it does not receive is your identity: the request arrives from ZeroLog.ai under one shared service identity, not from you at your address. The benefit is separation, not secrecy.
Is ZeroLog.ai completely anonymous?
No, and we would rather say so plainly. You hold an account with an email address, and paying for a subscription leaves its own trail.
What ZeroLog.ai gives you is a layer of separation between you and the underlying platform, and no permanent server-side chat history.
Where is my chat history kept?
In your browser, on the device you used. Nothing is kept on our side, which also means that clearing your browsing data for this site deletes your history.
In a private or incognito window the app says so and treats the session as temporary. You can export everything to a file at any time.
Can I use ZeroLog.ai on more than one device?
Yes, through a folder you already control. Point ZeroLog.ai at a folder in iCloud Drive, Dropbox or similar, and your history reconciles between your own devices through it.
That folder is yours. Your history moves through your storage, not through ours, and we never read it.
Which models can I use?
Claude Sonnet 5, Claude Opus 4.8, Claude Haiku 4.5, GPT-5.5 and DeepSeek V4 Flash, picked per conversation.
The lineup follows what our Swiss region serves, so it may change. One question goes to one model; ZeroLog.ai does not send it to several at once.
What happens to a PDF I attach?
Your browser reads the text out of it on your device. The file itself never leaves — no bytes are uploaded.
The extracted text travels with your turn like the rest of the conversation and is saved with it on your device. A scanned PDF with no text layer yields nothing: we do not read images for you.
Will my conversations be used to train a model?
Not by us. There is no permanent server-side copy for us to train on.
What a provider does with a request it receives is governed by that provider's own terms, which we do not control and will not paraphrase on their behalf.
Do you keep server logs at all?
Yes, operational ones — and they are built so that content cannot get into them. Every log line goes through a single seam that drops content fields, and the tests fail if content appears in one.
A request is correlated by a random identifier that is created for that request and never stored beside your account.
How does billing work?
Your usage is counted as token cost: how much you spent, never what you spent it on. The counter in your account is the same number we bill against.
Subscription billing is planned, with a tier that maps to a usage allowance. It may not be available at initial launch.
Your chats should belong to you, not to the platform.
Clear control over what is retained, and control that stays with you.