Just to clarify this, you can save any data you’ve added to files in clusters, but you can’t save which files a cluster contains. For Picard, clusters are only a temporary grouping intended to help Lookup be more accurate; they don’t have any meaning on their own. What does have meaning, though, is the releases (albums) Lookup – or Scan or manual assignment, but those don’t really benefit from clusters – matches your files with. Those are the groups Picard will wind up sorting your files back into the next time you open them, and those are (usually) what it uses to look up information from the database.
The one issue you’ll run into is if you have any mixtapes your friend gave you or any playlists you’ve put together with their own files. Picard does its matching by saving the album’s ID, but in order for it to have one to save, the release needs to be on MusicBrainz. We don’t have anywhere near as strict notability requirements as Wikipedia, but we do prefer an amount of distribution beyond just being passed around a circle of friends. Without that, there’s not really any good way to keep the files grouped; I’ve wound up just matching those tracks of mine to “standalone recordings” and only opening one personal album folder at a time. I can give you more information about that if you need it, but otherwise it’s pretty much unnecessary noise.