The Dual-Pane Review Screen is Docupath's central document validation environment, providing side-by-side access to extracted structured data and the original source document. This design enables rapid verification, allows reviewers to spot-check extraction accuracy, and supports intelligent field-to-document linking.
Understanding the interface components is essential for efficient document review and quality assurance workflows.
How It Works
The Dual-Pane interface divides the review screen into two complementary panels: extracted data on the left and the original document on the right. Both panels support interactive features and rich context indicators that help reviewers make informed decisions.
Left Panel: Structured Data
The left panel displays all data extracted by Docupath's AI in a structured format:
Dynamic Layout
Adapts automatically based on document type
Uses Docupath Internal Format (DIF) schema by default
Groups fields logically (e.g., party info, shipping details, line items)
Collapsible sections to manage screen real estate
Data Enrichment Indicators
Each field displays a contextual indicator showing its source and transformation status:
Indicator | Meaning | Example |
Verified Org Data - Updated | Field extracted and validated against org database | Invoice Number: org confirmed value |
Verified Org Data - Autofilled | Field populated from org reference data | Vendor Address: auto-filled from vendor master |
AI Enriched | Field supplemented with external data source | Tax Rate: 8.5% (enriched from tax database) |
AI Standardized | Field reformatted to standard format | Date: 2026-04-06 (standardized from "4/6/26") |
AI Transformed | Field value modified by transformation rule | Currency: USD (transformed from local currency) |
Interactive Field Inspection
Hover over info bubble icon (ⓘ) if it appears next to any field
Displays original extracted value vs. enriched value
Editable Fields
Click any field to enter edit mode
Save any changes made to documents
Manual override persists until document reprocessing
Change tracking available in audit log
Does not affect original document display
Right Panel: Original Document
The right panel displays the uploaded source document in its original format:
Document Display
Supports: PDF, images (JPG, PNG, TIFF), scanned documents
Full-resolution rendering
Page navigation controls (First, Previous, Next, Last)
Viewing Tools
Tool | Function | Shortcut |
Zoom In | Enlarge document view (up to 200%) |
|
Zoom Out | Reduce document view (down to 50%) |
|
Fit to Width | Auto-scale to panel width | Click button |
Fit to Page | Auto-scale entire page visible | Click button |
Rotate Right | Rotate page 90° (useful for sideways scans) | Click button |
Document View and Text View | Switch between original document and extracted text | Toggle button |
Select Mode | Enable bounding box selection | Toggle mode |
Bounding Box Visualization
When Select Mode enabled, draw bounding box over document elements
System updates selected field in left panel
Smart Field-to-Document Linking
Document → Field Direction
Select text or region in right panel document in Select Mode
Selected content auto-copies to active field in left panel
Overrides previous value without explicit save
Useful for rapid manual corrections
Field → Document Direction
Click on any extracted field in left panel
Corresponding source region highlights in right panel
Shows bounding box around source text
Helps validate extraction accuracy
Attachments
For documents that arrived by email, the viewer includes a View Attachments dropdown that opens the Attachments list — every file delivered in the same email, grouped together.
Attachments List
Each file listed by name (e.g., an invoice plus its supporting files)
Select a file to display it in the main viewer alongside the extracted data
Lets reviewers see the full set of paperwork that arrived together without leaving the page
Email Message Entry
Shown at the top of the Attachments list as an Email entry ("View full message")
Opens the original email body exactly as the sender wrote it; the viewer header shows "Email"
Provides context behind the documents (notes, instructions, forwarded threads)
Behavior
Display-only: the email message is never processed, read for data, or extracted — no effect on fields, totals, or approval
Renders both plain-text and richly formatted (Outlook/Gmail) emails
Links appear as text but are not clickable (intentional, for safety)
No Email entry appears if the email had no written body, or if the document was uploaded manually
Appears automatically — nothing to enable
Supported Configurations and Options
Document Type Adaptations
Different document types display tailored layouts. For example:
Invoice Layout
Party Information (Vendor, Buyer, Bill-To, Ship-To)
Invoice Metadata (Number, Date, PO Reference)
Line Items (Description, Quantity, Unit Price, Amount)
Totals (Subtotal, Tax, Shipping, Grand Total)
Purchase Order Layout
Requestor & Approver Information
Vendor & Ship-To Details
Order Metadata (Number, Date, Expected Delivery)
Line Items (SKU, Description, Quantity, Price)
Panel Sizing
Adjustable divider: Drag to resize left/right pane width
Persists per user session
Default: 50% left, 50% right (optimized for document viewing)
Other Technical Specifications
Dual-Pane Rendering Engine
Type: Client-side web components with server-side data support
Data Display: Dynamic form schema based on DIF
Performance: Sub-100ms field interactions
Docupath Internal Format (DIF)
Schema: XML-based structure defining document fields
Extensible: Custom fields supported via configuration
Versioned: DIF schema upgrades handled transparently
Validation: Field types enforced at display and save
Field Update Behavior
Manual edits trigger real-time validation
Server-side persistence after form blur
Optimistic UI updates with rollback on error
Concurrent edits: Last-write-wins (per document lock)
Enrichment Metadata Storage
Stored alongside field value
Includes: Original value, data source (e.g., trading party)
Bounding Box Coordinates
Format: Page-relative coordinates (not screen pixels)
Unit: Normalized percentage (0-100) for multi-resolution support
Precision: Sub-pixel accuracy for precise highlighting
Supported: PDF and image documents only (not XML)
Notes
Large Documents (100+ Pages)
Navigation may become slower on very large PDFs
Recommendation: Split into multiple documents or use search
Bounding box highlighting may lag on pages beyond 50
Zoom operations on large documents: 1-2 second latency
Scanned/Low-Resolution Documents
OCR accuracy may decrease on very poor quality scans
Bounding boxes may be less precise on heavily compressed images
Manual text selection recommended for accuracy
"View Extracted Text" toggle may show low confidence results
Multi-Page Field Spans
If field spans multiple pages, only first page highlighted
Bounding box shows only first occurrence
Manual review of subsequent pages recommended
Limitation: Technical constraint of current implementation
Concurrent Edit Conflicts
If two users view same document simultaneously, document opens in read-only mode for second user
Document lock prevents concurrent edits
Best Practice: Coordinate review assignments per document