The quickstarts in this section follow the same four steps:

4 steps, Python or Node.js

1Create a keyFree account
2InstallOne command
3RunChoose search, analyze, segment, or embed
4Read resultsClips, text, segments, or embeddings

Choose a task to open its quickstart. Each one gives you starter code to run and adapt.

Search for specific content in your videos using text or image queries.

Analyze videos and images

Analyze videos or images to create different types of text based on their content.

Segment videos

Detect timestamped segments in your videos and extract custom metadata for each one.

Create embeddings

Create embeddings for queries and for your media files: video, audio, images, and documents.

Search and embeddings use the Marengo model. Analysis and segmentation use the Pegasus model. You can use either model or both.

Prefer not to write code?

Try search, analysis, and segmentation on your own videos in the Playground, or start with the Jockey MCP with no code at all. Want Jockey from code? Use the Agents API.

For more in-depth information on each task, see the complete guides below:

Interactive notebooks

In addition to the quickstart guides, the interactive notebooks below provide complete, executable code for each task. Hosted on Google Colab, they allow you to experiment with the code directly, making them ideal for hands-on learning and customization.

Capability	Description	Try it
Search	Find specific moments in your videos.
Analyze	Analyze videos to generate text.
Segment	Segment videos into structured, timestamped data.
Embed	Create multimodal embeddings.