Intelligent Document Processing (IDP)

Last Updated: June 12, 2026

IDP is a Workflow step in Vasion Automate that uses AI to classify documents and extract structured field data automatically. When a document enters a workflow through a scan or upload trigger, the Intelligent Document Processing step analyzes the document, identifies its type, and extracts relevant data fields, eliminating the need for manual data entry.

How IDP Works

When the Intelligent Document Processing step runs in a workflow, it performs two operations in sequence:

  • Classification: The system identifies the document type, such as an invoice, based on its content.
  • Extraction: The system extracts structured field data from the document based on the fields defined in the selected document type.

IDP is powered by Amazon Web Services (AWS) Bedrock Data Automation. The step is available in the Advanced Steps section in the workflow builder and requires a document-based trigger, either scan or upload. Only one Intelligent Document Processing step is allowed per workflow.

This feature is in beta. Capabilities, configuration, and availability might change as development continues. Contact your Vasion representative with feedback to help shape this feature.

Requirements

Review these Requirements before you start.

In addition to the general requirements, an Intelligent Document Processing step requires the following:

  • The workflow must use a document-based trigger, either scan or upload.
  • You must have IDP permission assigned to your role.
  • This feature is available in the APAC, EMEA, and U.S. regions.

Document Types

The document type defines which fields the Intelligent Document Processing step extracts. There are two kinds:

  • Prebuilt document types: Vasion includes a standard invoice document type with preconfigured fields. You cannot modify prebuilt document types.
  • Custom document types: You can create your own document types by defining field names, data types, and optional guidance text. Guidance text helps the AI identify and validate each field more accurately. Custom document types are saved and reusable across workflows.

Each workflow is configured to process one document type at a time.

Confidence Thresholds & Review Routing

The Intelligent Document Processing step includes a configurable confidence threshold that ranges from 0% to 100% and defaults to 80%. After extraction the system scores each field based on how confident it is in the extracted data. One of the following happens:

  • If all fields meet or exceed the threshold, the document bypasses review and continues to the next workflow step automatically.
  • If any field falls below the threshold, the entire document is routed to an assigned reviewer.

The reviewer is a designated Vasion user that you select during workflow configuration. Reviewer assignment is required; the step cannot be saved without one.

Document Review

When a document is routed for review, the assigned reviewer receives an email notification and a task in their task queue. The task type is Document Review, and the task includes the document name, the person who initiated the workflow, and the date assigned. The reviewer opens the task by selecting Review.

The review interface has two panels:

  • Left panel: Shows the document with zoom, pan, and page navigation controls.
  • Right panel: Shows extracted fields in a vertical list.

The right panel has two tabs:

  • Items to Review: Shows only fields that fell below the confidence threshold. Editing a field removes it from this tab automatically.
  • All Items: Shows all extracted fields regardless of confidence score.

Selecting a field in the right panel highlights the corresponding area on the document in the left panel.

The reviewer can edit any field value. When a reviewer edits a field, the confidence score is removed and the field is marked User Modified. The reviewer can also complete empty fields, which are marked User Added. After saving, the reviewer can revert a modified field back to its original extracted value.

After the review is complete, the reviewer selects Submit to close the task and advance the document to the next workflow step. Selecting Close keeps the task open for later.

View Extracted Data

After a document is processed, its extracted data is available as read-only information on the File Information panel when viewing the file in Storage. Any user with permission to view the document can also view its extracted data.

If a document has not been processed with an Intelligent Document Processing step, the File Information panel shows a message explaining that the document can be processed through a workflow to extract data.

You can copy extracted field values by hovering over the field and selecting Copy.

Permissions

IDP access is controlled by the IDP permission, which you manage on the Roles and Permissions page.

Users with this permission can perform the following:

  • Add the Intelligent Document Processing step to workflows.
  • Configure the step, including custom document types and field definitions.
  • View extracted data for processed documents.
  • Export document data as a CSV, when available.

Next Steps

Refer to the following: