About Yataron conversation records
Explanation version: September 16, 2026 · Version 1
Why we keep records
YataLogic records conversation content and usage for administrator review to improve answers, question buttons, suggested links and the way the character is used. Conversation text is not used for advertising, sold or reproduced in public case studies. We do not send it to another AI or an external analytics service, or automatically register it as training data or knowledge.
What we do and do not record
We record accepted questions and final answers, their sequence, selected question buttons, input method, clicks on designated links, confirmation of display, technical failure reasons, date and time, whether the homepage or dedicated page was used, screen-width category, and versions of the interface and response policy. Screen width is not used to identify devices or count unique people.
Drafts, keystrokes, screen recordings, authentication tokens, detailed browser information and raw IP addresses are not stored in the analytics database. We also do not store the text of questions rejected by authentication or usage limits. Existing cost and security records and hosting-provider access logs are managed separately.
Handling of text and retention periods
Do not enter personal or confidential information. We use simple removal of recognizable email addresses and phone numbers from questions and answers prepared for storage, and do not store text suspected of containing secret keys or similar credentials. This cannot fully detect names, addresses or information that is confidential in context, and does not guarantee complete anonymization.
Question and answer text becomes unavailable through the administrator interface and CSV exports 30 days after acceptance and is deleted by a daily process. Individual conversation information and interaction records are retained for 90 days. Daily aggregates without text or individual IDs are retained for 365 days. Administrators can delete problematic conversation text and related records individually.
Copies may remain separately in hosting backups. A 30-day retention period in the active database is not the same as complete deletion of every copy. If data is restored, retention limits are applied before access resumes. We do not create our own long-term backups.
How this differs from response generation and browser storage
To generate answers, we send your input and the necessary recent conversation history and consultation summary to OpenAI. This is separate from YataLogic’s records described here. Cloudflare Turnstile, used to prevent automated submissions, receives connection and browser information but not conversation text. Each provider processes and retains data under its own policies; YataLogic does not control those retention periods.
Save summary in this browser on the dedicated page is optional and off by default. Its on/off setting is not a consent setting for operator-side records. Response generation and operator-side recording still occur when you do not save a browser summary. Conversations are not automatically carried between the homepage and dedicated page.
Deletion requests and other inquiries
Contact us with the approximate date and time of use and whether you used the homepage or dedicated page. You do not need to resend unnecessary personal or secret information or the entire conversation. Because we do not maintain long-term personal identification, it may not be possible to locate a record.
Communication or recording failures may prevent confirmation of some interactions. A missing response-display notification is not treated as proof that the response was not displayed or read.