Skip to main content

Document Review and Validation

Using Docupath to review and validate information captured from documents

Document Review and Validation is the human-in-the-loop stage where AI-extracted data is examined, corrected, and approved or rejected by authorised users. This capability sits at the centre of the Docupath document lifecycle - between ingestion/extraction and export - and is accessed through the Document Queue and the Dual-Pane Review Screen.

The module provides status-based queue management, configurable column views, data enrichment indicators, confidence colour-coding, and a structured action bar for document disposition.


How It Works

The Document Queue

The Document Queue is the primary workspace for managing documents across their lifecycle. It presents all documents in a tabular queue organised by status tabs - Pending, Validated, Approved, and Rejected - with each tab reflecting a distinct stage of the document lifecycle.

Top Panel Features

The top of the Review Screen provides navigation and filtering controls. Users can redirect to the Upload Screen, search documents by name or content, and apply filters by region, urgency, document type, and organisation. Active filters are highlighted with a red bubble indicator for visibility.

Tab View and Status Indicators

Each tab contains sub-statuses that provide granular visibility into document state:

  • Pending tab: Processing (AI still extracting), Pending (ready for review), Failed (unsupported format or corrupted file)

  • Validated tab: Documents reviewed by a Validator but awaiting final disposition

  • Approved tab: Approved, Sent (pushed to external system), Acknowledged (external system confirmed), Not Acknowledged (sent but unconfirmed)

  • Rejected tab: Documents rejected manually by reviewers or automatically by business rules, each with a required rejection reason

The Dual-Pane Review Screen

When a reviewer opens a document, the dual-pane layout presents AI-extracted data on the left and the original document on the right. This side-by-side design enables direct comparison between raw input and processed output.

Left Panel - Extracted Data

Displays structured data in Docupath Internal Format (DIF), with a dynamic layout that adapts to the document type being processed. Fields are annotated with enrichment indicators showing whether values were verified from organisation data, AI-enriched, AI-standardised, or AI-transformed. Confidence colour-coding provides at-a-glance quality assessment.

The left panel also Includes expandable fields and layouts such as line-item tables, which can be expanded to view more information at a glance.

Right Panel - Original Document

Shows the original uploaded document in its native format (PDF, image, XML, etc.) with page navigation, zoom, rotation, text extraction, and Select Mode for drawing bounding boxes to capture data directly from the document.

The right panel also includes a dropdown to view attachments if any are present.

Smart Field-to-Document Linking

Clicking an extracted field in the left panel highlights the corresponding section in the original document. Conversely, selecting text in the document copies it into the active field - enabling rapid correction without manual typing.

Data Enrichment Layers

During processing, Docupath applies multiple enrichment types to extracted data:

Enrichment Type

Description

Verified Org Data – Updated

Corrected using verified organisation settings

Verified Org Data – Autofilled

Populated from verified organisation settings

AI Enriched

Inferred contextually from document content

AI Standardized

Formats normalised (dates, currency, identifiers)

AI Transformed

Business rules and transformations applied

Footer Action Bar

At the bottom of the Review Rcreen, the Footer Action Bar provides the primary disposition controls:

  • Approve: Finalises the document and optionally routes it to external systems

  • Validate: Available only to Validator role users; moves the document to Validated status

  • Reject: Requires a rejection reason, which is logged for audit transparency

  • Close: Exits the document without applying changes

The footer also includes tab switching between the Input Tab (original document) and the Destination Tab (machine-readable output preview, e.g., XML).

Bulk Actions

For high-volume environments, the Document Queue supports bulk operations on selected documents: Delete, Download, Reprocess, and Review. Bulk review accelerates throughput for repetitive workflows.


Supported Configurations and Options

Configuration

Options

Column visibility

Show/hide columns (Document Number, Uploaded Time, Document Type, Primary Party, Document Date, Status, Action)

Column labels

DUP (duplicate), URG (urgent)

Tab views

Pending, Validated (add-on), Approved, Rejected

Filters

Regions, Documents, Organizations, Urgency, Document types, Status, Date, Validated by

Search

By document name or document content

Bulk actions

Delete, Download, Reprocess, Review

Validation lifecycle

Standard (Pending → Approved/Rejected) or Extended (Pending → Validated → Approved/Rejected)


Other Technical Specifications

Parameter

Detail

Maximum files for bulk review

50 documents at once

Supported original document formats in right panel

PDF, DOCX, DOC, XLSX, XLS, XML, JPG, PNG, TIFF, TIF, HEIC, HTML, TXT

Data presentation format

Docupath Internal Format (DIF)

Enrichment indicators

Hover-accessible info bubbles showing original vs. enriched values

Right panel features

Page navigation, zoom in/out, rotate, view extracted text, Select Mode with bounding boxes


Notes

  • Reprocessing a document resets all manual field edits - including manually assigned trading parties - and returns the document to Processing status

  • The DUP and URG column labels require their respective features to be active

  • Documents in Failed status (under the Pending tab) cannot be reviewed and require user action (re-upload or format correction)

  • Bulk review is limited to 50 documents per operation

  • Filter configurations are saved for each user and persist across browser sessions.

Did this answer your question?