How it works

Agent Number 10 runs one loop for each site.

1. Measure. Ask the AI engines the questions your buyers ask and record which answers cite or name you. Check where you rank in Google. Crawl your pages.
2. Find what to fix. Turn the audit into a fix plan: open issues grouped by type and ordered by severity, showing whether an exact change is available and how many pages one change fixes.
3. Fix. Some fixes are exact: the change is supplied word for word, ready to apply. Others are guided: you get the constraints and the page’s own facts, and write the change from those.
4. Verify. Check the live site to confirm the fix is really there. Page fixes are read from the live page and files such as robots.txt or the sitemap are fetched; site-wide issues, such as duplicate pages or redirects, are confirmed by the next audit. Then keep measuring to see whether the numbers moved.

A fix only counts while it still holds on the live site. If a later check finds it gone, it stops counting.

Two kinds of search, one project

AI answers and Google search are measured side by side because they feed each other. AI engines lean on pages that already rank and are easy to read, and the same technical problems hurt both. The Overview puts the two next to each other so you can see them move together, or not.