Overview
The Document Intelligence section provides a detailed overview of your document automation results for the chosen time period (e.g., Current Month to Date). This section is organized into three panels that allow users to assess extraction accuracy, processing volumes, and areas that require the most manual intervention.
- Documents Written To SoR: Tracks the volume of documents processed to your system of record and shows key accuracy metrics.
- Documents Dismissed: Reports on documents that were captured but not processed, such as duplicates or dismissed vendor submissions.
- Top 10 Fields with the Most Edits: Identifies which extracted fields most frequently required manual edits.
Each panel provides table and chart views, export functions, and up-to-date statistics critical for monitoring AP automation quality.
Figure 1. Document Intelligence Section Overview
Documents Written To SoR
This panel summarizes all documents the system has processed and written to the system of record (SoR) for the current period, as well as the accuracy with which these documents have been classified and extracted.
Figure 2. Documents Written To SoR – Table and Chart View
Table 1. Documents Written To SoR – Fields and Descriptions
| Field Name | Description |
|---|---|
| Documents Written to SoR | Documents written to the system of record (SoR) in the period, counted by write date. A document whose write raised a validation error in your system of record is not counted while that error is open. It is counted once the error is resolved, ignored or dismissed. |
| Documents Written Autonomously | Documents the agent wrote to the SoR with no user involvement. No extracted field was edited, no user review is recorded against the document, and the agent made the write rather than a user. |
| Documents Written with No Edits | Documents written to the SoR with none of the extracted header or line fields edited. A user may still have reviewed the document or made the write, so this count includes Documents Written Autonomously. |
| Documents Classified as Invoice | The number of these documents that were identified as invoices by Auditoria. |
| % Correct in Identifying Invoices | Percentage of written documents that were correctly identified as invoices. |
| Overall Accuracy (%) | Combined field extraction accuracy (both standard and extended) across all processed documents. |
| Standard Field Accuracy (%) | Accuracy for standard fields (e.g., Invoice Number, Total Amount, Invoice Date, Vendor Name, etc.). |
| Header Accuracy (%) | Accuracy for document header data (such as vendor name, invoice date). |
| Line Accuracy (%) | Accuracy for line-item/row-level data (such as quantities and line prices). |
Documents Written Autonomously and Documents Written with No Edits both exclude documents entered manually. The panel excludes credit memos from every count.
Actions:
- Toggle Views: Use the icons at the top right of the panel to switch between table and bar chart views.
- Hover for details: Move the pointer over any bar to show a tooltip with the period (e.g., August), the metric name (e.g., % Correct in Identifying Invoices), and the exact value (count or %). The bar is highlighted, values respect the current filters/date range, and on touch devices, you can tap to show/hide the tooltip.
- Export Data: Use the export icon to download datasets or visualizations in supported formats (CSV, PNG, JPG, PDF, SVG)
Documents Dismissed
This panel shows documents that were processed by SmartVendor but were later dismissed and not taken forward for business use—The agent dismisses duplicates by itself. A user dismisses the rest, either a document the agent classified as Others or a vendor bill. Monitoring these trends helps reduce lost data and surface recurring input or classification issues.
Figure 3. Documents Dismissed – Table and Chart View
Table 2. Documents Dismissed – Fields and Descriptions
| Field Name | Description |
|---|---|
| Count of Dismissed Documents | Documents dismissed and not written to the SoR. The figure is the sum of the four rows below it. Documents still sitting in Active Data are not counted. |
| Duplicates Identified by Auditoria and Automatically Dismissed | Number of documents auto-dismissed as duplicates by the system. |
| Documents Classified as Other by Auditoria and dismissed by user | Documents auto-classified as “Other” and manually dismissed. |
| Documents Classified as Vendor Bill by Auditoria and Changed to Other by User | Vendor bills reclassified as “Other” and dismissed by the user. |
| Documents Classified as Vendor Bill by Auditoria and Dismissed by User | Vendor bills directly dismissed by the user. |
Actions:
- Toggle Views: Switch between detailed table and summary chart views using the icons.
- Hover for details: Move the pointer over any bar to show a tooltip with the period (e.g., August), the dismissal category (e.g., Classified as Vendor Bill — User Dismissed), and the exact value (count). The bar is highlighted, values respect the current filters/date range, and on touch devices you can tap to show/hide the tooltip.
- Export Data: Use the export icon to download datasets or visualizations in supported formats (CSV, PNG, JPG, PDF, SVG)
Top 10 Fields with the Most Edits
This table identifies which extracted fields were most frequently edited manually after automation. Regular review helps users target improvements where the most manual intervention is needed.
Figure 4. Top 10 Fields with Most Edits
Table 3. Top 10 Fields with the Most Edits – Fields and Descriptions
| Field Name | Description |
|---|---|
| Field | Name of the invoice/document field edited. |
| Header/Line | Indicates whether the field is header-level or line-level. |
| Count of Edits | Number of times this field was edited in the given period. |
| Count of Total Fields | Total appearances of this field in processed documents. |
| % Edited | Percentage of total occurrences that were edited. |
Actions:
- Export Data: Use the export icon to download datasets.