IQ Bot and Document Automation v.27 release

Review what's new and changed, and the fixes and limitations in IQ Bot and Document Automation for the v.27 release.

Document Automation

What's new
Standard forms integrated with Document Automation

You can create and manage learning instances entirely within Document Automation. Previously, you had to create standard form learning instances in IQ Bot and then process them through the Connect Learning Instance feature in Document Automation. With standard forms integrated with Document Automation, you benefit by having an entire workflow consolidated into a single system.

Create custom models in Document Automation using Standard Forms.

Support for Azure AI Document Intelligence v3.0 models in Document Automation

Azure AI Document Intelligence v3.0 provides additional capabilities to the existing model creation, including options to select Template and Neural. While training a model, you can select either 2.1 or 3.0 version.

New Document Automation product licenses
Document Automation now offers the following licenses to enable users to process documents and track metrics for learning instances in production:
  • Document Automation Standard Forms (Number of pages): Required to create a model and process documents in a Standard form in Document Automation.
  • Document Automation 360 Extraction (Number of pages): Required to process documents in an Automation Anywhere (Pre-trained) model, including the User-defined document type.

Also, there are updates to license consumption tracking.

License tracking release note

Extract data with Google Vision OCR

If you are a user with a Cloud Control Room, you can process documents using Google Vision OCR technology by creating or updating a learning instance.

Improved table extraction in Automation Anywhere models

You can use validation feedback to train a learning instance to extract data from a cell that contains more than one field.

For example, if a product description column also includes item number, you can outline the item number in the Validation interface. When the learning instance processes subsequent documents, it will extract the item number and ignore the product description.

Use validation feedback to extract specific values in a table

Visualize performance metrics with a new Document Automation widget

You can view data visualizations on Validator user performance. This new widget is a horizontal bar chart that shows the weighted average time each Validator user spent performing manual validation per document.

Document Workspace dashboard

Document Automation supports regular expression pattern-based extraction

When configuring fields that the learning instance will extract, you can specify a regular expression (regex) pattern to help the extraction engine find the field. After entering the regex pattern you can test the expression using sample content within the Create Learning instance window. You can use regex pattern extraction for values that follow a specific pattern, such as a VAT number or a PO number.

Validation interface support for shortcut keys

Zoom in and out on documents in the Validation interface by holding the Alt key and scrolling with your mouse wheel.

What's changed
The following updates impact how the Control Room tracks license consumption:
  • The Control Room checks enough product licenses prior to deploying the bot.
  • The IQ Bot Pages license and the Standard Forms Pages license track documents that are processed through learning instances created in Automation 360 IQ Bot. This includes documents processed in the learning instance in Automation 360 IQ Bot and learning instances connected to Document Automation.
  • In the Licenses page, the Document AI license is called the Google Document AI (number of documents)
  • The Control Room also tracks documents processed in Google Document AI for users who brought their own license. Previously, the Control Room only tracked such documents for users who bought their licenses through Automation Anywhere.
You can now configure Oracle Database with on-premises and Linux installations of Document Automation.
When you import a learning instance created in previous versions of Document Automation, the check box to send validation feedback to the system is now not selected by default. You must edit the learning instance to select the check box.
Fixes
Build 16694: This build includes the following fixes (along with fixes from the previous builds):
Now users with either the Standard Forms for Document Automation or the Standard Forms (for IQ Bot) licenses can create a Standard Forms model in Document Automation.
Build 16685: This build includes the following fixes:
The Bot Agent now clears out the resources folder after running actions from the Document Extraction package.
When you import a learning instance created in a previous version of Document Automation, the Improve accuracy using validation check box is now unselected by default.
Document Automation now correctly tracks license consumption in all On-Premises installations, irrespective of the port number.
You can now add formula validation to learning instances with the user-defined document type.
Standard forms in Document Automation is now supported in a fresh installation or upgraded multi-tenant Control Room. Previously, Standard forms in Document Automation was only supported on a fresh installation of a multi-tenant Control Room.
When you refresh the Validator page, the Validator displays the last document you worked on. Previously after a page refresh, the Validator sometimes displayed the first document you had validated.
Bot Runner devices configured with a proxy can successfully run actions from the Document Extraction package.

Service Cloud case ID: 01856276, 01862174, 01911425

Limitations

Standard forms only support unique field extraction for any document. If there are fields with the same name on multiple pages, then the document must be trained with different tag names in the model creation tool.

For example, if there is a field named date on two different pages, then you must use date1 and date2 to make the tag names unique in the model creation tool for training.

You cannot use spaces in model names when you create models for Standard forms v3.0. If the model name contains any spaces, then an InvalidParameter error message is generated.

There is no average accuracy for standard forms v.3.0 from the Azure AI Document Intelligence service. As a result, the standard forms training displays the average accuracy as 0.00% for the v.3.0 service.

When fetching a Standard Forms model, an incorrect error message displays when no configurations are available in Document Automation.

The resizing functionality is not working properly for the model list table. As a result, you cannot resize the table header or column in Standard Forms.

The run analysis result does not display in standard forms when creating a v.3.0 model with multipages.

There is a default character limit of 100 characters for the table name, and 155 characters for the field name of the table. However, if you reconfigure the table, you can add more characters than the default limit.

The System-Identified Regions (SIRs) for checkboxes or radio buttons are not generated for the standard forms v.3.0 model.

Standard forms is not supported on Linux.

Any re-mapping or re-drawing in the System-Identified Regions (SIRs) in the Validator results in detection of an incorrect check box. However, you can enter the check box value as checked or unchecked after changing the auto-generated mapping.

When migrating a learning instance from one Control Room to another, the same user must export and import the .dw file.
If the learning instance extracts data from a multi-line field and transfers it to a CSV file, an extra space is added at the end of each line.
When you mouse over data points in the "Average validation time" and "Validator performance by number of documents" widgets, the correct values do not immediately display in the pop-up window.
Document Automation prevents you from creating a learning instance if you create a custom table field without providing a field name and field label, even if you have unselected the field.
If you click Test regular expression and replace the value in the Regular expression field, the field now has a limit of 50 characters.

Workaround: Click Cancel to leave the Test regular expression window and make the revisions in the Regular expression field.

The Table icon is not generated on the Form OCR Testing Tool (FOTT) for the Standard form v3.0 Model.

If a field name contains more than 100 characters, the scroll bar does not appear and you cannot see the value of the form field.

If a table name contains more than 50 characters, you cannot expand the table size. Additionally, the field name in the table does not display when you hover over it.

If you test an invalid Python script when configuring custom logic, the system returns an error message that contains the Python execution path.
Sometimes document extraction fails and one of the following error messages appear in the Automation Co-Pilot request status: Extraction bot failed. Please check activity/audit logs for more detail or Download bot failed. Please check activity/audit logs for more detail with this message in the audit log: Existing deployment in progress for this user session.

If this happens, restart the Bot Agent on the impacted device.

The Document Workspace dashboard sometimes shows incorrect page consumption metrics if the Control Room is installed On-Premises.
Documents cannot be processed with Google Vision OCR if your proxy is configured with a Proxy Auto-Configuration (PAC) file.
When you create or edit a Learning Instance, if a field name starts with a numerical value, then the formula validation for that field fails. This issue applies to form fields and table fields.
If you import a learning instance with the validation feedback option disabled, the validation feedback option becomes enabled and you must manually edit the learning instance to disable the option.
For Standard Forms, when a document with multiple pages is processed and enters into validation, one page is being duplicated in the Validatoreven when the System-Identified Regions (SIRs) in the Validator are correct.

IQ Bot

What's changed
If you are using an older version of Microsoft SQL Server that does not support TLS 1.2, then you must keep an IQ Bot database backup and update to a Microsoft SQL Server version that supports TLS 1.2.
IQ Bot now uses the latest version of Python (3.11.0) for custom logic validation. To use the latest version, you must upgrade from cx_Oracle 8.3 to python-oracledb. See Upgrading from cx_Oracle 8.3 to python-oracledb
Fixes
IQ Bot now performs post-processing on fields with Python custom logic that contain single quotation marks, such as print('Test').

In IQ Bot, when a document is either corrupted, deleted, or is not loading in Validator or Designer, a placeholder image is displayed. The applications will not be consistent until you upload new documents.

Service Cloud case ID: 00792976, 00835371

During the production process, you must not move the learning instances or bots from Staging to Production, or vice-versa.

Service Cloud case ID: 01808931, 00821541, 01253373, 00836954, 01271722

When you upload any invalid document in manual or custom groups, then the document gets processed. However, the System Identified Region (SIR) will not be generated for the unsupported pages.

Previously, when an invalid document was uploaded with a valid document and processed for extraction, the See extraction page kept loading.

Service Cloud case ID: 01786098

In Healthcheck API, the output shows the proper branch name convention, for example: RC-A360.27. Previously, improper health information was displayed when the branch name convention was heads/RC-A360.27.

If an error or exception is encountered during document validation, the error or exception is processed, and the files are displayed consistent in the output.

Service Cloud case ID: 00827940

In IQ Bot, if the RabbitMQ v3.8.17 and v3.8.18 is not working, then the log directory location can now be changed by updating the system environment variable as RABBITMQ_LOG_BASE. If this variable is provided manually with the required directory location, then the logs are created and updated in a new location. Previously, the configuration given for the log.dir variable in rabbitmq.conf file did not support RabbitMQ v3.8.17 and v3.8.18.

Service Cloud case ID: 01844995

The system-identified regions (SIRs) are now generated for manual groups when the PDFBox is enabled. Previously, when the PDFBox was enabled, during Learning Instance creation the SIRs for some vector PDFs were not generated for manual groups.

Service Cloud case ID: 01844223, 01849638

You cannot import an IQ Bot archive (IQBA) file when the designer for the same Learning Instance is open in another window. Previously, the import was failing because the LockedTimestamp field in the visiobotDetails table had a value at the time of IQBA export. This caused an error because the LockedTimestamp field was expecting a null value.

Service Cloud case ID: 01818474

Limitations
During the express installation upgrade of IQ Bot, an error message is shown in the installation steps. You must click OK to proceed further.
The maximum upload file size limit is set to 5 GB when importing and exporting files in the IQ Bot archive (IQBA) file. If the file size is greater than 5 GB, then you must delete unwanted groups to decrease the file size and migrate successfully.

If the IQ Bot archive (IQBA) file size is above 2,147,483,647 bits, then the File Storage does not upload the exported file from output folder to the desired location.

Sometimes, the PDFBox is unable to detect PDF (Portable Document Format) documents accurately. We recommend that you use Adobe Acrobat to convert the documents to PDF before you upload it to IQ Bot.

Updates to the interface

IQ Bot and Document Automation
The Create learning instance window now contains a drop-down menu to select an OCR provider.
Drop-down menu to select OCR
Extract data with Google Vision OCR