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

# Forward

> Forward a document from this pipeline to another pipeline for specialized processing.

<Note>This node is in **preview**. Its behavior may change in future releases.</Note>

The forward output hands a document off to another pipeline for further processing. It is a terminal node. Nothing connects after it within the current pipeline; the document's journey continues in the target pipeline, which then runs independently.

Use forward outputs to chain pipelines together when one pipeline is a good fit for one stage of processing and a different pipeline is better for the next stage.

<Warning>The target pipeline **must** contain a [Forward trigger](/nodes/forward-trigger) to receive the forwarded document. Pipelines without a forward trigger are not selectable.</Warning>

## Configuration

| Field               | Type            | Required | Description                                                                                                                             |
| ------------------- | --------------- | -------- | --------------------------------------------------------------------------------------------------------------------------------------- |
| **Target Pipeline** | pipeline picker | Yes      | The pipeline that receives the forwarded document. Only pipelines with a [Forward trigger](/nodes/forward-trigger) appear in the picker |

When this node is placed downstream of a [Classify action](/nodes/classify) in AI mode, the **Target Pipeline** picker is filtered to the target pipelines already chosen on the classify node, so each branch lines up with the pipeline it classifies for.

## How forwarding works

1. The current pipeline reaches the forward output node
2. The document is forwarded to the target pipeline
3. The target pipeline starts a new run triggered by its forward trigger
4. The two runs are recorded separately. You can open each one from the [Runs](/runs/monitoring) page

## Inputs and outputs

**Allowed inputs:** filter, split, classify, loop, extract, review, validation, transform, merge, parse, HTTP action, variable, store, reconcile, prompt.

**Output:** Terminal node, no downstream connections. The document is delivered to the target pipeline's forward trigger.

## Related

<CardGroup cols={2}>
  <Card title="Forward trigger" icon="share-from-square" href="/nodes/forward-trigger">
    Accept documents forwarded from another pipeline
  </Card>

  <Card title="Classify action" icon="tags" href="/nodes/classify">
    Classify pages and forward each one to a different target pipeline
  </Card>

  <Card title="Callback output" icon="webhook" href="/nodes/callback">
    Send results to an external webhook instead
  </Card>
</CardGroup>