The search bar lets you find footage across your entire AI-indexed library, by exact keyword or by meaning. Use it to locate clips by what's said, who appears, what objects are visible, or any combination of metadata, so you can discover content in seconds without manual logging or scrolling through files.
Find the search bar
The search bar sits in the top navigation bar and stays visible on every page, so you can search from anywhere in the platform.
Switch between keyword and semantic search
Outside a specific Collection, use the mode toggle in the search bar to choose how you search:
Keyword Mode: exact text matches across transcripts, titles, file names, custom metadata fields and MXT3 (Moments Lab's video understanding AI) generated descriptions. Use it when you know specific terms, names or phrases, especially with media files that follow strict naming conventions.
Semantic Mode: finds moments by meaning, even when you don't know the exact words. It combines textual and visual embeddings to match your query against MXT3-generated descriptions and the footage itself, so it can surface relevant shots that were never described in those exact terms.
Results from both modes are ranked by relevance, and each one includes a short explanation of why it matched: by keyword, semantic text or semantic visual similarity.
Search with labels
For more precise results: search using labels. Label search lets you search using tags, people and other structured metadata items applied to your media, created by MXT3 or through manual tagging. You can also scope a global search to one Collection by adding an in Collection tag to the search bar.
Use these prefixes to search by type:
@ (personality): type @ followed by a name to find a person from the thesaurus, for example @donald trump.
# (label): type # followed by a term to find a specific label.
Aa (plain text): search using plain text instead of a structured tag.
Combine labels with boolean logic
Combine with AND (default): add several labels side by side to find content that matches all of them. Use this search method when you want to find people who appear together in the same video or at the same moment within a video.
Combine with OR: drag one label onto another to find content that matches either one.
Exclude a label: double-click a label to exclude it from your search.
Combine labels and text
Pair a person label with free text to find moments when someone appears on screen and says or does something specific. For example, searching a person's label with a phrase finds moments when that person is on screen and the phrase appears in the transcript.
What's next
Need more help? Reach out to our Support team, we're happy to help.
Semantic Mode requires content processed by MXT3. Visual matching only covers content added after August 2025; older content needs to be reprocessed to support it.
UI not matching? Your platform may not be updated yet, let us know and we'll help.