Notes

How to find a file when you cannot remember what it is called

· 3 min read

You know the document exists. You know roughly what it said, who sent it, and that it was some time last spring. You do not know what it is called, and every search box on your computer wants a name.

This is the most common way of losing a file, and it is not a memory problem. Names are the one attribute of a document you never chose and never read.

Search what you remember instead

You almost always remember one of three things: a phrase that was in it, who it came from, or roughly when. Each of those is searchable if the right thing has been indexed.

A phrase from inside the document is the strongest. Both macOS and Windows index the contents of text documents and PDFs, and most people never try it because the search box looks like it wants a filename. Type a distinctive phrase and see what happens. Distinctive is the important word: search for an unusual pairing rather than a common one.

Who it came from usually means your email. If the file arrived as an attachment, the mail is a better index than your filesystem, because it carries the sender, the date and the thread.

Where this breaks down

Three cases defeat ordinary search, and between them they cover most of the files people actually lose.

  • Scanned documents and photographs. A PDF that is a picture of a page has no text to index unless something has read it.
  • Screenshots. Same problem, in the format most likely to contain the thing you half remember.
  • Anything you only remember the gist of. If what you recall is it was about the boundary with next door rather than a specific phrase, keyword search cannot help, because your memory is a summary and the index holds words.

Making the first two searchable

Both are fixed by reading the image. In Stash every photograph and scan goes through a read on the way in, which produces the text that is visible in it and a short description of what it shows. That text is indexed with everything else, so a screenshot of an error message is findable by the error message.

This is also why the naming matters. A file called Screenshot 2026-03-14 at 09.41.22.png contributes nothing to a search; one called Pricing page, three tiers contributes the words you are likely to try.

Making the third one searchable

The gist problem is different in kind. No amount of indexing helps if the words you remember are not the words in the document.

This is what Ask is for. You describe the thing the way you remember it, in a sentence, and it goes through your library and answers, showing the documents it used underneath. It is the difference between searching for boundary fence and asking what did the surveyor say about the fence.

The things it used are shown on purpose. An answer with no sources is a guess you cannot check, and for anything that matters you want to open the actual document rather than take a summary at its word.

A short procedure

  1. Try a distinctive phrase from inside the document, in your normal search. It works more often than people expect.
  2. Try your email for the sender or the thread.
  3. If it is a scan, a photo or a screenshot, stop searching for the name. Search for what is in the picture, if anything has read it.
  4. If you only remember the gist, describe it in a sentence and let something read the library for you.

The preventative version

Everything above is recovery. The reason files get lost is that they arrive with names that say nothing and go somewhere provisional, and both of those are fixable at the moment they arrive rather than a year later when you need the thing.

Stash does the filing

Drop anything in and it gets read, named for what is actually inside it, and put where it belongs. Ask for it back in plain words. A hundred things free.

Open Stash

Keep reading