# Making files searchable (indexing)

For the AI assistant to search and analyze a file, that file must be **indexed** — Albatross
extracts and prepares its content after upload. This happens automatically; no button to press.

Each file shows one of three states:

- **Indexing** — a spinner labeled "Indexing"; the file appears dimmed. The app checks
  automatically and clears this when done (seconds to minutes depending on size).
- **Searchable / Ready** — full brightness; the assistant can search it.
- **Not searchable** — a warning icon; the file is dimmed. **Hover the label for the specific
  reason.** Common causes:
  - A scanned PDF that is images-only with no extractable text ("No searchable content found").
  - An unsupported or corrupt file.
  - Certain media that couldn't be processed.

If a user asks the assistant about a file that's still indexing, the assistant will say it can't
search it yet and to wait for indexing to finish. Transient indexing failures are retried
automatically a few times; a file stuck at "Not searchable" usually means the content genuinely
can't be extracted (e.g. a scan with no text layer).