How does OCR search work?

When you add an image, scan, or PDF to a note, Harbor’s server runs OCR — optical character recognition — on it. It reads the text in the file and records where each word sits on the page.

That does two things for you:

• Search sees inside your files. A search for “furnace” finds the warranty PDF and the photo of the serial-number sticker — not just notes where you typed the word yourself.
• Matches are highlighted in place. Open the attachment from a search result and Harbor highlights the exact word right on the page, so you’re never left skimming a forty-page scan for the one line that matched.

There’s nothing to turn on and nothing extra to pay for. Capture normally — scan a page, snap a photo, drag in a PDF, or let a watched folder import a whole stack for you — and the indexing happens in the background on our servers. It even covers attachments your device hasn’t downloaded, so search stays complete without filling your disk.

The words OCR finds can do a second job if you ask for one. Press Suggest title on a scanned note and they’re what Harbor names it from — or switch that on for a watched folder and it does the same for every file the folder imports, so a stack of scans arrives titled rather than numbered. Both hand that text to an AI, which is the one thing here that goes further than OCR; the folder switch is off unless you turn it on.

One plain exception: encrypted notes. Harbor genuinely can’t read what you encrypt, so encrypted notes and their attachments are excluded from OCR and search. That’s the deliberate trade-off — lock down the sensitive things, keep everything else fully searchable. Details in What’s the trade-off when I encrypt a note?

More on search as a whole at Search & OCR.