AI output and data processing
What the AI does, what it cannot be trusted to do, and who processes your content.
Last updated 17 September 2026
AI output can be wrong
Analysis, themes, audiences, comparable titles, advertising copy and recommendations are generated by AI. They are suggestions drawn from your text and from research sources — not facts, not legal advice, and not a promise of results. Read everything before you use it.
Nothing goes public unreviewed
Marketing claims start unreviewed and must be approved by a person before they can be exported or published. Advertising campaigns are created paused. Emails only go to people with a recorded permission. Where a figure has never been measured, the page says so instead of showing a comforting zero.
What happens to your manuscript
The manuscript is stored privately. When you ask for an analysis, the text is read in sections and those sections are sent to the AI provider for that request only. It is not used to train public models, not shared with other workspaces and not published anywhere.
Who processes your content
- Hosting, database and file storage — running the service.
- AI provider — analysis, copy and images, at the moment you request them.
- Research and keyword providers you connect — to look up public web and search data.
- Payment provider — subscriptions and book orders. We never receive card numbers.
- Email provider — messages you choose to send, and the receipts we send you.
Records you can inspect
Each analysis finding keeps the excerpt it came from, each research result keeps its source link, and each administrator or billing action is written to an audit record. Where an integration is not configured, the page says "not configured" rather than showing it as connected.