How about setting it up so that Tags are included when Auto Classify is figuring out where items in the inbox need to be stored. For example when I scan in medical receipts, Devonthink can rarely figure out what folder the random receipts should go in. If DT would look at the tag I applied to an item, medical, then look at what folders also have a bunch of items tagged medical, but the item in that folder. It can be a little annoying when you scan in a manual, name the file with “manual” in the name, and tag it as a manual and DT still can’t figure out what folder the doc should go in (the Manual folder). Make sense?
DEVONthink classifies documents by comparing the content of the document to documents with similar content in the database, so the document name and/or tag are not factors in the classification. You can test this by creating a new, blank text file and name/tag it any way you desire and the See Also/Classify pane will be blank as there is no content to use for classification.
One option that may work for you is to turn on Group (gray) tags in your database (File menu>Database Properties…>database name and uncheck 'Exclude Groups From Tagging) to auto-classify documents based on tags. Then, if you are working with documents in the Inbox and tag one with the Group tag ‘Manual’, a replicant will be created in that group. You can then delete the document from the Inbox, or file it somewhere else if desired.
As documents can have multiple tags arched to a document with the same name, one additional suggestion I would make if you are going to use Group tags would be to delete any regular (blue) tags that have the same name as a Group (gray) tags.