Not every problem needs AI. Sorting and naming uploaded documents is one people reach for AI on, and they should not. I did it with a plain rule table that is faster, free, and never hallucinates. Here is how.

Six parts, and there is not a model in sight.

1. Read the name and a peek inside. The classifier looks at the file name and a little of the content to decide what it is.
2. Run an ordered rule table. Rules are checked in order, and the first one that matches wins. Order is the whole trick.
3. Rename to one convention. Every file gets a clean, consistent name, so a folder of chaos becomes a folder you can scan.
4. File it in the right place. The document lands in the correct folder automatically.
5. No AI, so no surprises. It is deterministic. The same file sorts the same way every time, and it costs nothing to run.
6. Add a rule by editing a row. When a new document type shows up, you add a line to the table. No developer.

Any business drowning in client uploads, accountants, agencies, can put this in front of the pile. Boring, reliable, free. The rule table above is the entire build, and it will never send you a bill.

From a bank platform build. Everything described here is something I actually run; nothing on this bench is theoretical.