Privacy

    Your library is yours.

    This page describes what Synthara collects, why it collects it, and what it never does with it. Plain language, no dark patterns.

    What is collected

    When you sign up, Synthara stores your email address and an account identifier from the authentication provider. When you upload a document, the file contents and a vector index of its passages are stored so questions about it can be answered later. When you ask a question, the question and the response are kept so the conversation is there when you come back.

    Standard request metadata is logged for one week to debug outages. That log is purged on a rolling basis.

    What never happens

    Your documents and conversations are not used to train models. They are not sold, licensed, or shared with third parties for marketing. They are not used to improve a product that other users will benefit from. Your library stays scoped to your account.

    Who can see your data

    Only you. In the rare case where a debug requires it, you will be asked for explicit permission and only the specific record needed will be accessed. Every such access is logged.

    Subprocessors

    To run Synthara: a managed Postgres host stores account data, a managed object store keeps the original file bytes, a vector database powers semantic search, and one large language model provider handles the answer step. Each is bound by a data processing agreement that prohibits training on your data.

    Your controls

    You can delete any document, any conversation, or your entire account from the Workspace and Library pages. Deletion is permanent. The file bytes, the vector index, and the message history are purged within seven days.

    Questions

    Privacy questions can go to c9ivira@gmail.com.