1. Information you provide
Depending on the feature you use, you may provide:
- past writing samples used to estimate your communication style;
- incoming messages that you want help answering;
- prompts and response-generation requests; and
- candidate replies submitted for personalized ranking.
Avoid submitting sensitive personal information or content you do not have permission to process.
2. How the current system uses information
The FastAPI backend processes submitted text in memory to create a style profile, rank candidates, or prepare a generation request. It limits the working history to the most recent 100 messages. Conversations, incoming messages, generated replies, and response ratings are not intentionally persisted. Writing samples you explicitly save through the Your voice editor are stored in the project's Supabase PostgreSQL database, associated with your account, until you remove them or request account deletion.
Account authentication is stored separately in a PostgreSQL database. That storage is used for account details, linked Google provider records, and one-way password hashes for email accounts. Login state is carried in an encrypted session cookie. Saved writing samples are accessible through authenticated server routes for their owning account; they are not exposed through the browser's Supabase Data API. Account self-service deletion is not available yet; contact the project owner with a deletion request.
3. Operational logging
The backend logs request identifiers, endpoint paths, response status codes, and request latency for reliability and debugging. It is designed not to log raw writing history, incoming-message text, candidate text, generated replies, API keys, or provider response bodies. New frontend code does not log submitted text.
4. External AI provider
When you request AI generation, relevant writing history, the incoming message, and generation instructions may be transmitted to the external AI provider configured by the operator. That provider processes the request under its own terms and privacy practices. PickyTalker does not make claims about the provider's retention periods or other practices; check the configured provider's current policy before using generation with personal or confidential material.
Style profiling and ranking are performed by the PickyTalker backend. The browser does not contact the external AI provider directly.
5. Accounts, cookies, and analytics
PickyTalker uses Auth.js with Google OAuth or an email-and-password account. Google provides basic profile information such as your name, email address, and profile image. For email accounts, PickyTalker stores a one-way password hash rather than the password itself. Auth.js uses an encrypted, essential session cookie to keep you signed in. PickyTalker does not add analytics, advertising trackers, or non-essential cookies, so it does not display a cookie-consent banner for the current feature set.
6. Data security and choices
Reasonable technical measures are used to limit exposure, including backend-only provider credentials and privacy-safe application logging. No system is completely secure, so submit only information you are comfortable processing through the service. You can choose not to use AI generation or not to provide writing samples. Open Your voice to review or change saved samples. Remove every sample and save to stop using that writing history for future replies. Messages you receive and AI-generated replies are never automatically added to your writing samples.
7. Future changes
Privacy practices may change if later versions introduce conversation persistence, analytics, additional sign-in methods, or additional AI providers. Before those features are launched, this policy and relevant user controls should be updated to explain what is collected, why it is used, how long it is kept, and how it can be deleted or changed.
8. Contact
Privacy questions can be sent to abhanage06@gmail.com.