Privacy
What Atlas processes and what it keeps.
When you run a tool
The text and files you provide are processed only to complete the workflow. AI-assisted workflows send extracted content to Atlas’s AI provider; deterministic workflows such as exact document comparison and spreadsheet reconciliation do not. Atlas does not intentionally save raw inputs or generated results in its own database. The provider may process submitted content under its applicable terms, so avoid material you are not permitted to share.
When you request a tool
Your request is reviewed for safety, support, and overlap. A created workflow is unlisted by default. Atlas stores its declarative configuration and a one-way hash of the normalized request, not executable code. The raw request is not intentionally retained after processing.
Search and usage analytics
Atlas uses Google Analytics and its own aggregate product analytics to understand page visits, tool runs, completion or error events, downloads, exports, feedback, and which results are selected. Google Analytics may receive standard browser, device, approximate-location, and cookie or similar measurement data. Atlas does not send your tool inputs, uploaded file contents, generated results, or search phrases to Google Analytics. Search phrases may be normalized and stored separately by Atlas with aggregate result counts; obvious email addresses, URLs, and long numbers are removed before storage. Atlas does not need your name or an account to use these features.
Abuse prevention
Atlas uses a one-way identifier derived from an IP address and a private server secret to enforce request limits. It stores only the rate-limit bucket identifier, count, and expiration—not the raw IP address in that rate-limit record.
Limits
No internet service can promise absolute privacy or security. Atlas minimizes what it keeps, limits input size, and keeps provider credentials on the server. There is no Atlas account or saved-workspace feature in this version.