> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bowerlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Hybrid semantic search

> Find anything in your workspace — even if you don't remember the exact words you used.

Bower's search finds anything in your workspace — even if you don't remember the exact words you used.

## How it works

Bower uses hybrid semantic search: a combination of keyword matching and vector embedding. In practice, this means:

* **Exact terms** are found reliably — searching for "pH 8.1" returns notes that contain that value.
* **Meaning-based queries** also work — "water alkalinity last Tuesday" will surface relevant notes even if those exact words don't appear together in any note.
* **Scientific synonyms** are handled — "sodium chloride" and "NaCl" return the same results.

Search covers everything in your workspace: voice transcripts, text extracted from photos, PDF extractions, imported documents, note content, and attachment metadata.

## How to search

<Steps>
  <Step title="Open search">
    Press **Cmd+K** (Mac) or **Ctrl+K** (Windows/Linux), or click the search bar at the top of the desktop app.
  </Step>

  <Step title="Type your query">
    Use natural language or keywords.
  </Step>

  <Step title="Browse results">
    Results appear ranked by relevance, with the matching excerpt shown in context.
  </Step>

  <Step title="Open a result">
    Click any result to open the note.
  </Step>
</Steps>

## Tips for better results

* **Be specific about time** — "field samples from February" narrows results more than "field samples."
* **Use your own terminology** — Bower learns from what you capture, so your specific shorthand and abbreviations become findable over time.
* **Search from Bird** — for questions that need synthesis rather than retrieval, asking Bird directly is often more useful than a keyword search.

## What's indexed

Everything captured or imported into Bower is indexed for search immediately — there's no delay between capture and searchability. This includes the full text of transcripts, photo extractions, and document content, as well as titles, tags, and metadata.