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

# Export Annotations

> Export completed annotations as datasets (JSON/CSV) for fine-tuning, evaluation, or analysis.

## What it is

Export lets you turn annotation results from a queue into a structured dataset you can use for fine-tuning, evaluation, or offline analysis. You can export directly into a FutureAGI dataset or download as JSON/CSV.

## Export to Dataset

<Steps>
  <Step title="Open Export to Dataset">
    Open queue detail and click the **Export to Dataset** button in the header.
  </Step>

  <Step title="Choose a target dataset">
    Create a **new dataset** by entering a name, or select an **existing dataset** from the dropdown.
  </Step>

  <Step title="Filter by status">
    Optionally filter by item status. By default, only completed items are included.
  </Step>

  <Step title="Run the export">
    Click **Export**. The annotations are written as rows in the target dataset with all label values as columns.
  </Step>
</Steps>

## Export as JSON/CSV

<Steps>
  <Step title="Open the Export menu">
    Open queue detail and click the **Export** button. Choose your format -- **JSON** or **CSV**.
  </Step>

  <Step title="Filter by status">
    Optionally filter by item status to include only the records you need.
  </Step>

  <Step title="Download the file">
    Click **Download**. The file is generated and saved to your local machine.
  </Step>
</Steps>

## Export data structure

Each exported record contains the following fields:

| Field        | Description                                           |
| ------------ | ----------------------------------------------------- |
| item\_id     | Queue item ID                                         |
| source\_type | Type of annotated source (trace, span, session, etc.) |
| source\_id   | ID of the annotated entity                            |
| status       | Item status (completed, skipped, etc.)                |
| annotations  | Array of label values with annotator info             |
| notes        | Annotator notes (if any)                              |

## Use cases for exported data

* **Fine-tuning** -- Use annotated traces as training data for model improvement.
* **Evaluation datasets** -- Create golden datasets for automated eval pipelines.
* **Quality reports** -- Analyze annotation patterns and model failure modes offline.
* **Model comparison** -- Compare model outputs across annotated dimensions.

<Note>
  Export to Dataset creates a full FutureAGI dataset that you can use with all dataset features including experiments, evaluations, and prompt management.
</Note>

<Tip>
  For programmatic export, use the [Queues API](/docs/api/annotations/queues/export) or the [SDK export methods](/product/annotations/sdk/python).
</Tip>

## Next steps

<CardGroup cols={3}>
  <Card title="Analytics & Agreement" icon="chart-pie" href="/product/annotations/features/analytics">
    Review annotation progress and agreement before exporting.
  </Card>

  <Card title="Dataset Overview" icon="database" href="/docs/dataset">
    Learn about FutureAGI datasets and what you can do with exported data.
  </Card>

  <Card title="Queues API" icon="code" href="/docs/api/annotations/queues/export">
    Export annotations programmatically via the REST API.
  </Card>
</CardGroup>