semantic search

semantic search ranks documents by meaning instead of exact words. it uses stored document embeddings and same-origin session auth.

updated may 3, 2026 documents no images
quick answer

use semantic search when the user remembers the idea but not the exact phrase.

use this for

  • finding related drafts
  • searching large accounts
  • explaining why results can show without exact keyword matches

steps

  1. open the dashboard.
  2. type a query of three or more characters.
  3. wait briefly while the query debounces.
  4. review ranked results.
  5. open the matching document.

details to know

  • embeddings are stored separately from documents.
  • existing documents can be backfilled.
  • autosave avoids re-embedding when the relevant input has not changed enough.

limits and edge cases

  • empty documents are not useful for embedding.
  • if the embeddings provider fails, local keyword search still works.
ready to use the product instead of reading about it? open app