buffera Documentation
Back to siteDownload Buffera
Documentation

Search by words and meaning

Find an exact phrase, text in an image or a related idea.

Start with words

Ordinary search works without downloading a model. Every query word must appear in the search index; letter case does not matter. Type, app and date filters narrow the results.

Text in saved PNG/TIFF images is recognized in the background by Apple Vision on your Mac. Once recognition finishes, those words become searchable. Selecting the image still copies the image.

Add semantic search

  1. Open Settings → AI.
  2. In Semantic search, choose Download model and enable. EmbeddingGemma 2 uses about 310 MB.
  3. Wait for the relevant part of your history to be indexed. Settings show the progress.
  4. Enable Also search by meaning in the history window.

Word matches appear first, followed by related clips. All filters still apply. If the model is unavailable, ordinary search remains available.

Models, battery and search limits

Semantic search requires Apple Silicon. The model downloads separately and runs locally. Background indexing pauses on battery by default; Continue now resumes it. Search over already indexed clips stays available.