Understand on-device processing and privacy
Understand the files used for suggestions, the local processing, and the controls that change your tags.
Last updated: 2026-09-11
Quick answer
Little Tagger’s suggestion pipeline extracts available file information and uses local analysis, including Apple’s on-device model when ready. It does not upload those documents to a remote tagging service. The input comes from workspace files, not only the highlighted selection. Opening the AI inspector can start that processing automatically.
What information is used
The suggestion feature can use filenames, containing-folder context, available document text, text recognized in supported images, dates, and existing or recently used tag names. Extraction is limited by supported formats, file size, page counts, and the amount of information the model can accept.
The result is a set of proposed tags for the workspace. It is not a complete reading of every document or a guarantee that every image or scanned page has been understood.
Choose the analysis scope
Before opening View > Inspector > AI, keep only the intended input files in the workspace. The generation path receives the imported workspace list. Highlighting a subset, entering a search, or filtering the displayed rows does not establish a smaller private analysis boundary.
Use Remove Selected Files to remove particular workspace items, or Clear Files to start a separate input set. Those controls do not delete the original files.
Granting a folder under Locations and adding files to the workspace are different operations. Locations are used for tag discovery and management. Do not treat workspace filters as permission controls for either feature.
Know when processing and changes happen
| Action | Effect |
|---|---|
| Open the AI inspector with files and no current generated result | Can start suggestion analysis automatically. |
| Click Generate tag suggestions or use Actions > AI Tag Suggestions | Requests a new analysis. |
| Double-click a suggestion | Prepares the tag; a new name is also created in the library. A selected preset can be edited by this selection. |
| Click Apply | Runs the prepared tagging action on the current target files, according to execution timing. |
| Drag a suggestion into the workspace | Directly adds that tag to the current target files. It is not just a preview. |
To do a manual tagging task without requesting file-content suggestions, stay in the Tags inspector rather than opening AI or running its generation command.
Local processing is not the same as no network activity
The document-analysis and tag-generation pipeline described here runs locally. Preparing Apple's models can require a download. App Store purchases and access verification can require a connection. Opening the website or contacting support is also separate from local suggestion generation.
The local pipeline description is not a statement that every Apple Intelligence feature, every website service, or every part of macOS works offline.
A blocklist is not content redaction
Blocked Tag Suggestions removes matching names from the suggestions returned to you. It does not strip matching words from source documents before the local model sees them. Use the workspace input controls when a file must not be analyzed.
Clearing the workspace also does not erase tags already written to files, your saved library, or copies and backups maintained elsewhere.
For the website and contact-data policy, read Privacy. For questions about the app's processing of a particular file type, contact support with a description first; do not send a confidential document as the initial example.