Citeroot is built around AI systems — both as the thing we measure and as a tool we use. This policy explains our commitments.
#What we measure
We collect outputs from third-party AI systems for prompts you define. We store these outputs as collected and label them with the engine, date and context. They are not statements by Citeroot and may be wrong.
#Where we use AI
- Detection in recorded answers (whether your brand is named, whether an answer reads as a recommendation) uses simple, explainable rules rather than a model, and shows the evidence passage.
- Briefs and insights (beta) are drafted from facts computed from your project's data. If the operator has configured an Anthropic key, Claude writes the draft; otherwise a template does. Drafts are labelled as drafts.
#Commitments
- No training on your data. We don't use Customer Data to train models. Prompts and brand names are sent to the AI providers listed under subprocessors to collect answers; each provider's API terms govern what they do with them.
- Human approval. Citeroot produces reports and drafts for review. It doesn't publish content, contact third parties or change your website.
- Transparency. AI-written outputs are labelled as drafts, and insights show the facts they were written from.
- Scoped access. Drafts are written only from the data in the project they belong to.
#Limits
AI-generated outputs can be inaccurate or incomplete. Review before acting. Don't use outputs for decisions with legal or similarly significant effects on individuals.
#Our crawler
We identify our user agent as CiterootBot, honour robots.txt and only fetch pages in response to a user's request. See CiterootBot.
#Contact
Questions or concerns: privacy@citeroot.app.