Privacy
Message texts and visitor data are kept for 15 days, the context of a complaint (up to the last 500 messages) for 30 days, and the staff action log for 90 days. The service has no advertising trackers. Source: the “Retention periods” section below, version of September 19, 2026.
Version of September 19, 2026. This document describes the technical pilot of the service.
What data is processed
To run the chat we use a random session identifier, IP address, browser and operating system string, browser language, the selected search parameters, messages and complaint materials. The device model is determined only to the extent the browser reports it; no separate digital fingerprint is created. Geolocation, contacts, IMEI, MAC, camera and microphone are not requested.
Retention periods
| Data | Retention period |
|---|---|
| Message texts and visitor data | 15 days |
| Complaint context (up to the last 500 messages), reason and technical details | 30 days from the date of filing |
| Staff action log | 90 days |
| Technical links between records, blocking decisions and service metadata in backups | up to 14 additional days |
| History of the conversation with the AI waiter (in memory) | until the search ends or you disconnect; cleared after 10 minutes without new messages |
Message texts and visitor data are kept for 15 days. The context of a complaint — up to the last 500 messages, the reason and technical details — is stored separately for 30 days from the date of filing. The staff action log is kept for 90 days. The keys for texts and technical records are stored separately from SQL backups, stop being used when they expire and are deleted automatically. A SQL backup does not extend the ability to decrypt this information.
Technical links between records, blocking decisions and service metadata may remain in backups for up to 14 additional days. After a restore, expired records are purged before the service is opened. Backups of the archive keys are not created.
Who has access
Your conversation partner sees only the messages of your conversation. A moderator sees complaint materials and the technical list of active sessions. The super-administrator can additionally read any available correspondence, including spot checks without complaints, and appoints moderators. Viewing actions are recorded. End-to-end encryption is not used: the server processes messages and stores an encrypted archive.
Third-party services and cookies
Hosting is provided by Vultr. Protection against automated sign-ins is provided by Cloudflare Turnstile: the browser contacts this service and the server verifies the result. The necessary cookies contain random visitor and staff tokens, are protected with HttpOnly and are sent over HTTPS. There are no advertising trackers. Conversations are not stored in the browser's localStorage.
Conversation with the AI waiter
If the model is connected and you choose to talk to the bot, the server sends Google Gemini your messages to the bot and a limited context of that conversation. Conversations with people, IP addresses, profile data and the visitor identifier are not included in the model request. The bot's history is kept in memory until the search ends or you disconnect, or is cleared after 10 minutes without new messages at the next cleanup. It is not written to the archive of messages between people. Processing on Google's side is governed by the Gemini API terms; it also depends, among other things, on the pricing plan used. Do not share personal or confidential information with the bot.
Inquiries
To report a violation, use the “Report” button in the conversation. Before the public advertising launch, the owner must add the operator's details and a contact for data-related inquiries.