Privacy

Two things on this site process data: the early access form, and the documentation assistant in the corner. This page describes exactly what each one does, because that is the whole of it.

The early access form

• The email address you enter. Required — it is what the list is for.
• Your name, if you fill it in. Optional.
• Which form you submitted, your IP address and browser user-agent, and the time. These come with the request and are kept to block abuse of the form.

Used for sending you launch updates and early access details for FisServer, and nothing else. No newsletter you did not ask for, no third party, no advertising, no resale of the list.

The documentation assistant

Questions you type into the chat widget, and the answers given, are stored. Rather than claim the arrangement is privacy-friendly, here is each choice and the principle it answers to, so you can judge it yourself.

• What is kept — the question, the answer, which model produced it, how many tokens it cost and how long it took. Nothing else. (Data minimisation: only what the purpose below actually needs.)
• What is not kept — your IP address and your browser user-agent. The signup form keeps both to block abuse; the assistant does not need them, so it does not have them.(Data minimisation again: the question is what people ask, not who asked.)
• Nothing identifies you — no cookie, no account, no tracking identifier. Questions from one visit are grouped by a random value generated in your browser and thrown away when you reload the page, so it cannot link two visits together. This is also why there is still no cookie banner on this site: there is nothing to consent to.(Purpose limitation: grouping a conversation, not recognising a person.)
• Why it is kept — to find out which questions come up often, so the documentation and this site can answer them properly. Not for profiling, advertising, resale, or training a model. (Legitimate interest, GDPR art. 6(1)(f): improving my own documentation, weighed against data that identifies nobody.)
• How long — twelve months, after which an exchange is deleted. The deletion is done by the code that writes each new one, not by a scheduled job somebody has to remember, so the limit holds on its own. (Storage limitation: a defined horizon rather than "as long as it is useful".)

One thing this does not protect you from: if you type something personal into the box, that text is what gets stored. Please treat it as a public question rather than a private message.

Answering a question sends it, together with the project documentation, to OpenAI in the United States, which generates the answer as a processor under its data processing terms and does not use data sent through its API to train its models. That is a transfer outside the EEA, covered by the standard contractual clauses in those terms.

Analytics and cookies

There are none. No analytics script, no tracking pixel, no cookie banner — because there is nothing to consent to. Fonts are served from this domain rather than a third-party CDN.

Where it is stored

Both in a private hosted database (Turso, acting as processor); signups also exist as a copy in the notification email sent to the site owner. Nobody else has access to either. Signup records are kept until FisServer launches and the launch announcement has gone out, and are deleted after that; stored conversations are deleted after twelve months, as described above.

Your rights

Under the GDPR you can ask what is held about you, have it corrected, or have it erased. Ask through the contact form at fiscet.it and it will be handled — deletion is one row removed from one table, so there is no reason for it to take long. For a stored conversation, quote enough of it to identify it: since nothing links it to you, that is the only way it can be found.

The data controller is Christian Zanchetta, reachable through the same form.