Existing content
Find which of your existing articles should link to each other - based on real semantic closeness, not keyword overlap.
The article content is turned into a semantic vector (an embedding), then matched against all of your own approved and published articles to find the closest ones in meaning. Articles in the same topic cluster get a ranking bonus. The result is a list of suggestions for you to pick from and insert yourself - nothing is ever auto-inserted into your content.
Enter the URL of the article you are optimizing. The tool matches against your own library, so it needs an account with approved or published articles in it - the more you have, the better the suggestions. A brand new site with nothing to match against will not have suggestions yet.
Internal linking is the cheapest way to pass strength from articles that are performing to ones that are not, and it is the step most sites skip once a post is published. An article nothing points to is hard for both readers and search engines to find.