Kaizendex DocsOpen the app

Files & Photos

Every file you've added as a table — what it is, what the AI read in it, where it's used — with columns you choose, a full-text reader, and corrections.

Files & Photos is the table of every file you've added to Kaizendex — pictures, PDFs, documents, spreadsheets, voice notes. Each row is one file; each column is one fact about it, including everything the AI worked out when it read the file.

Opening it

Open Settings and choose Files & Photos — or go straight to /files. On desktop, Settings is the gear next to your profile in the sidebar footer; on your phone, tap More in the bottom bar, then your profile at the bottom of the drawer. You can also search for "Files" from the search bar (⌘K / Ctrl K).

What each column means

These columns show by default:

  • File — the name you uploaded it with (a pasted picture shows "Photo"), with its pixel size underneath.
  • Type — PDF, JPEG, Word, Excel, and so on.
  • Size — how much space it takes.
  • Added — when it entered Kaizendex.
  • Used in — the entries and pages that contain this file. Click one to jump straight to it. Clicking anywhere else on the row opens the first of them.
  • Description — the short "what is this" line, written by the AI when you added the file or by you (see Files get a description you can edit).
  • Processing — where the file stands with the AI reader: Pending (queued), Processing, Ready, Partial (readable text but with a declared gap — hover to see what was skipped, e.g. "5 pages without a text layer"), or Failed (nothing usable; hover for the reason). Beside it, the coverage in one glance — "37/40 pages", "2 sheets · 120 rows", "2:30 of speech" — and "corrected" once you've fixed the text yourself.

Every other fact is a column you can turn on with the Columns button (grouped by where the fact comes from):

  • File — Dimensions, Duration, Checksum, Updated.
  • Source — Source URL, Source, Source ID, Published (when the source was published), Imported (when it came in), Language, Transcript source (manual captions, auto captions, speech recognition, or an upload).
  • AI — Described by (AI / you), Described (when), Description error, Text — the full extracted text or transcript.
  • Processing — Processing error, Coverage, Extractor, Model, Processed (when), Content version, Corrected (when you overrode the text).

Facts about the file and its processing are read-only. Only two fields take your edits — Description and Text — and they are marked "editable" in the Columns menu.

Choosing, ordering, sorting and filtering columns

Click Columns to show or hide any column and move it left or right. Your choice is saved to your account, so the table looks the same on your other devices. Click a column header to sort by it (again to flip, a third time to clear). Type in the filter box to narrow the list to rows whose visible columns contain your text, and pick a status to see only, say, the failed files.

Reading a long value

A description, an error, or a file's full text shows as a one-line preview in its cell. Click it to open the reader: the complete value, with Find in text (jump between matches), Copy, and a monospace toggle. For a file's text there is also a Segments view — a rail of where each piece came from (page 7, slide 3, sheet rows 12–40, 14:20–14:50 of a recording) that scrolls straight to it.

The reader is the same one you get on an entry page or in a table when a text property is long — look for the expand glyph beside the value.

Fixing the text, and processing again

If the AI misread a scan or misheard a name, open the file's text (the row's ⋯ menu → Read text) and click Correct. Save, and your version becomes the truth: the assistant reads it from then on, search updates, and the reader shows Corrected by you. Reprocessing the file later keeps your correction unless you choose Reprocess and discard my correction.

Reprocess file (in the row's ⋯ menu, or in the reader) asks the AI to read the original file again — useful after a failure, or once a better reader ships. The original bytes are always kept.

Check Processing and Coverage to see what was actually read. A partial file can still contain readable pages. Reprocessing uses the current limits, so it does not guarantee that every omitted page becomes available.

You can also ask the assistant to correct a file's text or process it again. Give it the correction you want; it preserves the rest of the text. Processing again keeps your corrections unless you explicitly ask to discard them. A queued request means the work is waiting to run; the file's Processing column shows when it finishes. If a save cannot be confirmed, the assistant says so instead of repeating it automatically.

Editing the Description works the same way: open it from the row's ⋯ menu or click the description cell, edit, save. See Uploading files and what the AI can read for what gets read from each kind of file.

On your phone

The phone shows the same table as a list: each row carries the columns you chose as short lines, and tapping a row opens a details sheet with every fact (hidden columns are dimmed), plus Open page, Read text, Edit description and Reprocess. Sort and Columns sit in the header. Long values open the same reader, with Share in place of Copy.

Where files come from

You don't upload to this page directly — files appear here when you add them to an entry. Paste, drag, or use /Image in any note (see Images in the body), attach a file in a chat, or set a photo on a Property, and it shows up in Files & Photos automatically.

Next