Scout, the AI agent.
Scout is the agent inside CiteHawk. It reads your visibility data and answers questions about it in plain language, so you can interrogate your numbers without knowing which surface holds them.
What it is
Scout has access to the same workspace data the rest of the product renders: scores, rankings, prompts, sources, recommendations, and the collected AI answers underneath them. Ask it a question and it goes and reads before it answers.
What to ask it
Scout is most useful for questions that would otherwise mean opening four surfaces and holding the answer in your head:
- Which prompts did I lose this week, and to whom?
- Which cited sources do not list us yet?
- What changed since the last collection, and why?
- What should I do first?
How it is grounded
Scout answers from your collected data and points at the evidence behind what it says. That grounding is the whole design: an agent that summarized confidently without showing its working would be exactly the problem CiteHawk exists to measure.
What it will not do
Scout reads; it does not change your workspace configuration on your behalf. Editing prompts, competitors, and settings stays a deliberate act you perform in the product.
If you want an agent that reads your CiteHawk data from outside the product, in Claude or ChatGPT or your own tooling, that is the MCP server rather than Scout.
