Unlike keyword search, Spote’s semantic search understands what you mean — not just what you typed. You can describe an idea, ask a question, or use completely different words from what you originally wrote, and Spote will still find the right notes.Documentation Index
Fetch the complete documentation index at: https://docs.spote.cloud/llms.txt
Use this file to discover all available pages before exploring further.
How it works
When you save a note, Spote builds a single text string from your note’s bucket, title, tags, and body, then generates a vector embedding of that combined text. The embedding captures the meaning of the note as a point in high-dimensional space. When you search, Spote embeds your query the same way and finds the notes whose embeddings are most similar. Results are ranked by semantic relevance, not by how recently you created the note.Semantic search indexes your note’s bucket, title, tags, and text together — so the more context you add, the better the results.
What this means for you
- Search for “machine learning decision” and find notes about “AI-powered recommendations” — even if you never used those exact words.
- Search by feel, not by memory. You don’t need to recall the precise phrase you used when you wrote the note.
- Works well even with hundreds of notes. Results are ranked by relevance so the most useful notes surface first.
Similar notes
While editing a note, Spote surfaces related notes in a collapsible panel below the editor. These are found using the same semantic similarity — Spote compares the current note’s embedding against all your other notes and shows the closest matches. This is useful for connecting ideas you’d forgotten were related, or for finding older notes to link or reference.Via API and MCP
If you integrate with Spote programmatically, semantic search is available in two ways: REST APIsearch_notes tool does the same thing from an AI agent such as Claude Desktop. Results include each note’s title, bucket, tags, text, and relevance score.