> ## Documentation Index
> Fetch the complete documentation index at: https://docs.1archive.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Search

> 1Archive has three search modes (AI Image Search, Transcript, and File Name), each designed for a different way to locate media across your sources.

Search in 1Archive spans your entire catalog simultaneously. Whether a file lives on a drive sitting under your desk or one that hasn't been connected in six months, 1Archive can find it. Because the catalog persists independently of physical drive connections, you don't need to plug in every drive before searching. If the drive was ever scanned, its files are findable.

There are three search modes, each suited to a different kind of query. You switch between them using the mode selector next to the search bar.

## AI Image Search

AI Image Search lets you describe what you're looking for in plain language or with an example concept, and 1Archive returns visually and semantically similar images and video frames, even if those files have no relevant filename or metadata.

This works because every image and video frame in a scanned source can be analyzed with **AI image indexing** during the scan process. The AI analyzes what each image actually depicts, its subject matter, mood, colors, and composition, rather than its filename or tags. When you type a search query, 1Archive matches your description to this visual analysis to surface the most relevant files.

What this means in practice:

* Searching `golden hour portrait outdoors` finds close-up photos taken in warm evening light, even if the files are named `DSC_4821.jpg`.
* Searching `aerial city at night` surfaces drone footage and cityscapes without any manual tagging.
* The results are ranked by visual and conceptual similarity, not alphabetical order or exact keyword matches.

<Note>
  AI Image Search only works on sources where **AI image indexing** was enabled
  during the scan. If a source was scanned without this option, its files will
  not appear in AI Image Search results. You can rescan a source to enable it.
</Note>

## Transcript Search

Transcript Search looks inside the spoken audio of your video files. When a source is scanned with transcript generation enabled, 1Archive transcribes the audio track of each video. Transcript Search then queries those transcriptions and returns clips where your search term appears in the dialogue or narration.

Results surface the specific video file and, when available, jump to the timecode where the spoken phrase occurs, so you don't have to scrub through an entire interview to find the one line you need.

<Tip>
  Transcript Search is especially powerful for interview footage, documentary
  shoots, and any project where spoken content needs to be retrieved quickly.
</Tip>

## File Name Search

File Name Search is a straightforward keyword match against the names of files and folders in your catalog. It returns any file whose name contains the characters you typed, regardless of file type or source.

This mode is useful when you already know the filename or naming convention you used (for example, searching `BTS_2024` to find all behind-the-scenes files from a particular year, or `_FINAL` to locate deliverable versions).

## Filtering Results

Regardless of which search mode you're using, you can narrow results with two filters:

<CardGroup cols={2}>
  <Card title="Filter by Source" icon="hard-drive">
    Limit results to files from one or more specific drives or source groups. Useful when you know which drive a file is on, or when you want to scope a search to a single project's media.
  </Card>

  <Card title="Filter by Date Range" icon="calendar">
    Restrict results to files created or modified within a specific date window. Set a start date, an end date, or both to zero in on a particular shoot day or time period.
  </Card>
</CardGroup>

<Info>
  Date filters apply to the file's creation or modification date as recorded in
  its metadata, not the date it was scanned into 1Archive.
</Info>