Citations and the advisor

Citations

In grounded mode (Starter and Pro) the assistants search the web before answering and their answers carry the sources they used. Every citation is stored with the run: URL, domain and the position it appeared in. The “Where the AIs look” card aggregates them by domain across your prompts — the review sites, directories, comparison pages and publications the assistants trust for your market.

This is the part of the metric you can act on: get covered on the domains that already shape the answers and the grounded score follows. Citations also feed the advisor below.

The advisor

Starter and Pro projects get the GEO advisor: for the prompts where your brand is absent or ranked low, an agent reads your brand profile, those prompts with the brands that win them, and the domains grounded answers cite, and produces three to five concrete recommendations — the comparison page to publish, the directory to get listed on, the review site to pursue, the llms.txt to add. Evidence-based: every recommendation names the prompts, competitors or cited domains that motivated it, with an effort estimate.

The advisor runs on demand from the dashboard (one run at a time per project) and its output is kept with the date, so you can measure what a recommendation did.