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

# Your first agentic editing export

> Export clips from Jumper using an AI agent in a few simple steps

export const ChatStatus = ({children}) => {
  return <div className="flex items-center gap-3 w-full py-1">
      <div className="flex-1 border-t border-gray-200 dark:border-white/10" />
      <span className="text-xs text-muted-foreground shrink-0 flex items-center gap-1">
        {children}
        <Icon icon="chevron-right" size={10} className="opacity-50" />
      </span>
      <div className="flex-1 border-t border-gray-200 dark:border-white/10" />
    </div>;
};

export const ChatResponse = ({children}) => {
  return <div className="self-start w-full max-w-[85%] min-w-0 flex flex-col gap-1">
      <span className="text-xs pl-2 font-semibold opacity-50">AGENT</span>
      <div className="rounded-xl rounded-bl-sm px-2 py-3 text-sm text-gray-800 dark:text-gray-100">
        <div className="prose prose-sm dark:prose-invert max-w-none overflow-x-auto">
          {children}
        </div>
      </div>
    </div>;
};

export const ChatMessage = ({children}) => {
  return <div className="self-end flex flex-col gap-1 max-w-[75%]">
      <span className="text-xs pl-2 font-semibold opacity-50">USER</span>
      <div className="rounded-xl rounded-br-sm dark:border dark:border-gray-500 px-4 py-2.5 text-sm bg-gray-100 dark:bg-white/20 dark:text-white">
        {children}
      </div>
    </div>;
};

export const Chat = ({children}) => {
  return <div className="not-prose chat-interface p-4 mb-14 border border-gray-200 dark:border-none dark:bg-white/10 rounded-lg">
      <div className="flex flex-col gap-3 my-2">{children}</div>
    </div>;
};

This tutorial walks you through your first agentic editing workflow: asking an AI agent to search for footage, review its plan, and export clips you can drop straight into your NLE.

For background on what agentic editing is, see [Agentic editing](/core-concepts/agentic-editing).

## Before you begin

* **Jumper** is running with [media analyzed](/core-concepts/processing)
* An **AI agent** (Claude or Codex) is [connected](/interface/settings-tab#ai-agent)

<Warning>
  Claude Cowork currently does not reliably pick up the Jumper integration. For this tutorial, use Claude Code Desktop, Claude Code CLI, or Codex.
</Warning>

<Steps>
  <Step title="Give the agent a search query">
    Open your AI agent (Claude Code Desktop, Claude Code CLI, or Codex) and describe what you're looking for in plain language. For example:

    <Chat>
      <ChatMessage>
        okay use Jumper to search in <code>\~/Downloads/jumper test movies/</code>, I want a varied sequence with calm landscape shots and other nature related topics. Lets aim for 20 clips
      </ChatMessage>
    </Chat>

    Tell the agent which media to search if it matters (e.g. a specific folder path).
  </Step>

  <Step title="Review and approve the plan">
    The agent will run some checks, and then come back with a plan: which media it will search, the query it will run, and what it will export.

    <Chat>
      <ChatMessage>
        okay use Jumper to search in <code>\~/Downloads/jumper test movies/</code>, I want a varied sequence with calm landscape shots and other nature related topics. Lets aim for 20 clips
      </ChatMessage>

      <ChatStatus>
        Worked for 10s
      </ChatStatus>

      <ChatResponse>
        Here's my plan:

        Run multiple visual searches with varied queries across the folder:

        * Calm water, ocean waves, beach
        * Mountains, valleys
        * Desert, dunes
        * Sunrise, golden hour
        * Village, countryside, pastoral
        * Forests, trees, greenery
        * Sky, clouds

        Collect \~20 clips by running several searches, then deduplicating and picking the strongest results.

        * Export as timeline XML or export individual clips to a folder.

        Question: Do you want the sequence as timeline XML or individual clips to a folder?
      </ChatResponse>

      <ChatMessage>
        lets export the clips to a folder
      </ChatMessage>
    </Chat>

    Review the plan and approve it when you're ready. The agent will then execute the search and export steps.

    In this case, we'll ask the agent to export the clips to a folder.
  </Step>

  <Step title="The agent exports clips to a folder">
    The agent uses Jumper to search your analyzed media, finds matching clips, and exports them as individual files to a folder you specified (or a default location).

    You don't need to click through Jumper yourself, the agent handles the search and export.

    <Chat>
      <ChatMessage>
        lets export the clips to a folder
      </ChatMessage>

      <ChatStatus>
        Worked for 32s
      </ChatStatus>

      <ChatResponse>
        All 20 clips have been exported as individual MP4 files to:

        **`~/Downloads/jumper test movies/landscape_clips`**

        **Folder structure:**

        | Subfolder    | Clips   |
        | ------------ | ------- |
        | `ocean/`     | 4 clips |
        | `mountains/` | 4 clips |
        | `desert/`    | 4 clips |
        | `sunrise/`   | 4 clips |
        | `village/`   | 4 clips |

        Each file is a trimmed clip you can import into Premiere and arrange however you like.
      </ChatResponse>
    </Chat>
  </Step>

  <Step title="Import the clips into your NLE">
    Open the folder with the exported clips and drag it into Premiere Pro, Final Cut Pro, DaVinci Resolve, or Avid Media Composer. The clips are ready to use.

    <img src="https://mintcdn.com/jumper/WefRMX_VnmgEBFbr/images/agent_clips_premiere.jpeg?fit=max&auto=format&n=WefRMX_VnmgEBFbr&q=85&s=470c70215ad20bb3bb98910522fb03f7" alt="agentic_editing_import.png" width="3104" height="1836" data-path="images/agent_clips_premiere.jpeg" />
  </Step>
</Steps>

## Timeline round-trip

Once you are comfortable exporting clips, try a round-trip workflow: the agent reads your active timeline in Premiere Pro, DaVinci Resolve, or Avid Media Composer, searches your analyzed media, and sends an updated timeline back.

<Steps>
  <Step title="Open the sequence in your NLE">
    Make sure the timeline you want to edit is active in Premiere Pro, DaVinci Resolve, or Avid Media Composer.
  </Step>

  <Step title="Ask the agent to read and update the timeline">
    Describe the edit you want in plain language. For example:

    <Chat>
      <ChatMessage>
        Use Jumper to read my current Premiere timeline, find better b-roll for each clip from <code>\~/day1\_footage/</code>, and update the timeline with the strongest matches.
      </ChatMessage>
    </Chat>
  </Step>

  <Step title="Review in your NLE">
    When the agent finishes, review the updated sequence in your NLE before locking the cut.
  </Step>
</Steps>

For background on what the agent can access, see [Agentic editing](/core-concepts/agentic-editing#timeline-import-and-export).

## Resulting sequence

Here's the resulting sequence from the above workflow. 20 clips exported by the agent and dropped into Premiere:

<div style={{ position: "relative", paddingTop: "56.25%" }}>
  <iframe src="https://customer-5b7nirmpkyszbexu.cloudflarestream.com/800c52560f614deaba569dc618453f8c/iframe?poster=https%3A%2F%2Fcustomer-5b7nirmpkyszbexu.cloudflarestream.com%2F800c52560f614deaba569dc618453f8c%2Fthumbnails%2Fthumbnail.jpg%3Ftime%3D%26height%3D600&autoplay=true&loop=true&muted=true" loading="lazy" style={{ position: "absolute", top: 0, left: 0, height: "100%", width: "100%", border: "none" }} allow="accelerometer; gyroscope; autoplay; encrypted-media; picture-in-picture;" allowFullScreen={true} title="Nature shots sequence exported via agentic editing" />
</div>

## What's next?

* Ask the agent to search for people (make sure you have faces detected and named in the [People tab](/interface/people-tab))
* Ask the agent to search in transcriptions to cut the best parts of an interview
* Generate a b-roll sequence from a specific voiceover track
* Ask the agent to export a sequence (Premiere XML, FCPXML) instead of individual files
* Ask the agent to read your active NLE timeline, swap clips based on search results, and send the updated timeline back
* Run multiple agent tasks in parallel for different projects

For more on what agents can do, see [Agentic editing](/core-concepts/agentic-editing).