UiPath Documentation
ixp
latest
false
IXP in Flow user guide
  • Overview
  • Model building
    • Overview
    • Managing Extract nodes
    • Uploading sample documents​
    • Reviewing predictions
    • Validating predicted values
    • Model configuration
    • Bring your own subscription (BYOS)
  • Model evaluation
  • Model deployment
  • Consuming models
  • Developer
  • Frequently asked questions

Managing Extract nodes

Add, configure, and remove Extract nodes in UiPath Maestro Flow.

The Extract node pulls structured fields out of a document, such as an invoice, receipt, form, or contract, using an IXP extraction model, and returns the result for downstream nodes to act on.

Prerequisites​

Before you add an Extract node, make sure that your tenant has both the IXP and Document Understanding services enabled. The Extract node depends on both services, even when you only use IXP models.

Adding an Extract node to a flow​

To add an Extract node to a flow, proceed as follows:

  1. Open your flow in Maestro Flow.
  2. Select Add node, then Document, then Extract.
  3. Select one of the following options from the node palette:
    • Extract document - Create a new node and, implicitly, a new model.
    • In this solution - Select an existing node from the dropdown list.
    • Published - Select from the dropdown list a deployed IXP model, such as a model deployed from IXP Unstructured and Complex Documents or from the Extract node of another flow. Each model appears as an individual node.

Result: The node is added to the Flow canvas, bound to the model you selected. At this point, you can connect the Extract node to the previous node in your flow.

Configuring the Extract node​

When you hover over an Extract node on the Flow canvas, a toolbar appears. You can select one of the following options to configure the node:

  • Edit - Opens the node properties, where you configure the inputs, error handling, and advanced settings of the node. For details, check Configuring the node in Properties.
  • Open full view - Opens the Definition tab, where you configure the extraction model and the taxonomy of the node. For details, check Configuring the node in Definition.

Configuring the node in Properties​

To configure the node in Properties, proceed as follows:

  1. On the Flow canvas, hover over the Extract node.
  2. In the toolbar that appears, select Edit.

Result: The node properties open in a window. The Properties section in the middle contains the settings that configure the Extract node. The Input panel on the left and the Output panel on the right are available for every node in Flow. For details, check Input and Output panels.

Properties​

The Properties section contains the following tabs:

  • Parameters
  • Error handling
  • Advanced

Parameters​
Note:

In case of validation errors, a number is displayed on the Parameters tab.

To configure the fields within the Parameters tab, proceed as follows:

  • File input:

    • File - The file to extract data from. Bind it to a file produced earlier in the flow, such as the output of a trigger or of a previous node. Use Insert to open the variable picker, where you can search for variables. The picker lists the outputs of the previous nodes under $vars, and the flow metadata, such as FolderKey, InstanceId, and ProcessKey, under $metadata.

    • Page range - Optionally, add the pages to process, for example 1-5. Leave it empty to process the whole document. Use the value type T icon in the field to choose between:

      • Fixed value - Enter a value directly.
      • Expression - Enter a JavaScript expression with IntelliSense.

      You can also use Insert to add a variable.

  • Definition - A summary of the extraction model of the node:
    • The extraction model and its version, together with the evaluation metrics. A node that has no evaluation run shows Not yet evaluated.
    • Annotated docs - The number of annotated documents out of the total number of documents.
    • Upload - Upload documents to annotate.
    • Annotate - Open the annotation view. This option is unavailable until you add fields to the taxonomy.
    • Run evaluation - Run an evaluation of the extraction model.
    • Edit - Open the Definition tab of the Extract document panel. For details, check Configuring the node in Definition.
    • Model config - A read-only summary of the model configuration: Extraction model, Extraction strategy, Language support, Layout parsing, and Attribution.
    • Taxonomy - The number of fields in the taxonomy. If the taxonomy has no fields, select Add fields to open the Definition tab with a new field group.
Error handling​

The Error handling tab contains the Enable error handling option. When you enable it, the node gets an error output handle that catches and handles failures, and the node output includes an error object with the code, message, detail, category, and status keys.

Advanced​

The Advanced tab contains the following settings:

  • ID - The ID of the node, for example extractDocument2. To change it, select the edit icon.
  • Icon color - The color of the node icon on the canvas.
  • Update variable - Assigns values to variables when the node runs. Select Add variable update, and then configure the following settings:
    • Variable - The variable to update. Select an existing variable, or select Create new variable.
    • Expression - The value to assign. Use Insert to add a variable. To reference a variable in an expression, use $vars.{name}.
Input and Output panels​

The Input and Output panels are available for every node in Flow. They show data from the flow, and you do not configure the Extract node in them.

Input​

The Input panel lists the previous nodes in the flow. You can view the input in the following formats:

  • Schema - The output keys of each previous node and their types. For a new node, the keys have no values.
  • JSON - The input in JSON format. For a new node, this view is empty.
Output​

The Output panel shows the output of the previous run of the node, with the version, type, result, taxonomy, and metadata keys. If the node has not run yet, the keys have no values. You can view the output in the Schema, JSON, or Extraction result format.

The dropdown list selects which response the node returns:

  • Live - The real response from the node.
  • Static mock - A value you define, returned every time.
  • Generated - A response that an LLM generates dynamically.

Configuring the node in Definition​

To configure the node in the Definition tab, proceed as follows:

  1. On the Flow canvas, hover over the Extract node.
  2. In the toolbar that appears, select Open full view.

Result: The Extract document panel opens the Definition tab.

The Definition tab contains the following sections:

  • Evaluation summary - The extraction model, its version and evaluation metrics. A node that has no evaluation run shows Not yet evaluated. For details, check Node health score. The section also contains the following options:

    • Annotated documents - The number of annotated documents out of the total number of documents.
    • Upload - Uploads documents to annotate.
    • Annotate document - Opens the annotation view. This option is unavailable until you add fields to the taxonomy.

  • Model configuration - The extraction model and its advanced settings. For details, check Model configuration.

  • Taxonomy - The fields that the node extracts from each document:

    • Overall extraction instructions - Add context that is relevant across the whole taxonomy.
    • Taxonomy tab - Add and manage field groups and fields. For a new node, select Field group to add the first field group.
    • Field types tab - Add and manage the pre-configured and custom field types.
    • Preview documents - Open a document preview next to the taxonomy.
    • Import/Export - Import or export the taxonomy. For details, check Importing and exporting taxonomies.

    For details, check Managing the taxonomy.

When you change the taxonomy, make sure you select Save.

Removing an Extract node​

To remove an Extract node from a flow, proceed as follows:

  1. On the Flow canvas, hover over the Extract node.
  2. In the toolbar that appears, select the ellipsis icon.
  3. Select Delete. Alternatively, select the Extract node on the Flow canvas, and press Backspace.

Result: The Extract node is removed from the Flow canvas.

  • Extract - The node reference, including how to read the extraction result, missing fields, confidence scores, and table fields.
  • Document processing - The document processing nodes available on the Flow canvas.

Was this page helpful?

Connect

Need help? Support

Want to learn? UiPath Academy

Have questions? UiPath Forum

Stay updated