Document Intelligence – Test group setup

Note

This feature is in preview, and the related documentation is subject to changes.

Document Intelligence – Test group setup enables users to create test groups in Microsoft Dynamics 365 Finance & Supply Chain Management from external specification documents, such as Word and PDF files. Instead of manually reading the document and entering each test line in the system, users can upload a supported file and let the application extract the relevant information, propose matching tests available in Microsoft Dynamics 365 Finance & Supply Chain Management, and prepare the data for review before import.

The feature is designed to reduce manual effort, improve consistency in test group setup, and support a controlled, traceable import process. The AI application does not directly create or approve a test group in . Instead, it proposes values in a staging area, where users must review and, if necessary, correct the information before importing it into the Test groups form.

When a document is uploaded, the system stores the file in Azure Blob Storage, triggers the Document Intelligence application, extracts the relevant quality-test information, performs semantic matching against the tests available in Microsoft Dynamics 365 Finance & Supply Chain Management, and returns the proposed results to the Document Intelligence staging form for review.

This capability is especially useful when test group definitions are maintained outside the ERP system and need to be translated into structured setup data with reduced risk of interpretation errors.

Pre-requisites

Before using this feature, ensure that the required setup is completed and that the relevant master data exists in Microsoft Dynamics 365 Finance & Supply Chain Management.

1. Follow the steps in the “Installation Guide for Document Intelligence – Test group setup” document. See Installation guides.

2. Configure Document Intelligence parameters (System administration > Setup > Document Intelligence parameters) by filling in the following fields:

   • Container Name: Azure Blob container where documents will be uploaded (refer to the Installation Guide above for additional details).

   • AI Agent Endpoint URL: Endpoint of the AI processing service (refer to the Installation Guide above for additional details).

   • Key Vault Certificate Name: Certificate used for authentication (refer to the Installation Guide above for additional details).

   • Minimum confidence score: Threshold for acceptable AI matches. This determines whether a proposed test is automatically accepted or flagged for review.

     Example: if the Minimum confidence score is set to 0.70, lines with a lower confidence score are automatically marked with the Exclude checkbox in the staging form. These lines are not imported unless a user reviews the proposal, corrects it if necessary, and clears the Exclude checkbox before import.

3. Verify Tests master data (Life Sciences Quality Control & Assurance > Setup > Quality control > Tests) and ensure that:

   • All tests expected in the imported document are available in the system and correctly configured.

   • The Description field is meaningful and accurate, because the AI matching compares the document text against the Test and Description fields in .

4. Verify Test variables for tests of type Option (Life Sciences Quality Control & Assurance > Setup > Quality control > Test variables) and ensure that:

   • A Variable exists for each test of type Option, with the Variable value equal to the Test.

   • Each of those variables has a Not verified outcome available, which is used as the default outcome for imported Option tests.

Test group upload and AI processing

Follow these steps to upload a document, review the AI proposal, and create a new test group in .

1. Open the Test groups form (Inventory management > Setup > Quality control > Test groups or Life Sciences Quality Control & Assurance > Setup > Quality control > Test groups).

2. In the Action Pane, click File upload.

3. In the Upload document dialog, select a supported file in .pdf or .docx format and confirm the upload. After the upload succeeds, the document is stored in Azure Blob Storage, and  sends a request to the Document Intelligence application to start processing the file.

   1. The file is uploaded to the Azure Blob Storage container configured in Document Intelligence parameters.

   2. The Document Intelligence application reads the uploaded document, extracts the relevant data, and performs semantic matching against the tests configured in .

   3. The proposed results are returned to  in JSON format and stored in the staging tables for user review.

4. Monitor the progress in the Document Intelligence staging form, which can be opened from the Test groups. Each header record represents one uploaded document and includes the Document ID, File name, Created date and time, Status, and Created by fields. Common processing statuses are:

   • Upload succeeded: the document was successfully uploaded to Azure Blob Storage.

   • AI request succeeded:  successfully notified the Document Intelligence application.

   • AI response succeeded: the AI response was received, and the staging lines were created successfully.

5. If processing fails, the status helps identify the point of failure. Examples include Upload failed, AI request failed, Response parsing failed, and Import failed. Use the Log button in the staging form to review the detailed processing history and related messages.

Review the AI response in the staging form

When the header status is AI response succeeded, the extracted data is available for review. Users must validate the proposal before importing it into the Test groups form.

Each extracted quality test is shown as a line in the Document Intelligence staging form. The line includes information from the AI response and fields that users can update before import.

• Overview tab:

  • Document text: the raw text extracted from the document, describing the quality checks to be performed.

  • Proposed test: the best-matching Test (i.e., the test among F&O available records getting the highest confidence score) found via semantic analysis.

  • Confidence score: a value between 0 and 1, indicating how certain the AI is about the proposed test selection, with 1 (100%) meaning complete certainty.

  • Reasoning: a description of the logic that led to the test match.

  • Exclude checkbox: automatically enabled when the confidence score for the line falls below the Minimum confidence score threshold specified in the Document Intelligence parameters.

  • Updated test: blank by default.

• Test tab: displays the proposed test data under the Proposed values section and provides the Updated values section for manual corrections.

Review each line carefully before import. If the proposed data is not correct, use the editable fields to adjust it as described below.

• If the Proposed test is correct but the proposed specifications are not, enter the correct values in the Updated values.

  Note

  If you fill in at least one field in the Updated values section, you must ensure that all other relevant fields in the same section are also completed correctly, because during import the system uses the Updated values section whenever at least one field in that section has been filled in.

• If the Proposed test is incorrect, select the appropriate value in Updated test. If the updated test has a different type than the proposed test, also review and complete the relevant fields in Updated values.

  Note

  • Fields available in Updated values depend on the type of the selected test. For tests of type Fraction or Integer, users can update result-range and measurement values. For tests of type Option, users can update the variable and default outcome.

  • If you fill in at least one field in the Updated values section, you must ensure that all other relevant fields in the same section are also completed correctly, because during import the system uses the Updated values section whenever at least one field in that section has been filled in.

• If a line should not be imported, select the Exclude check box. Excluded lines are ignored when the test group is created.

• If the confidence score is low, perform an additional review of the proposed test and proposed values. Low-confidence lines may already be automatically excluded, depending on the parameter configuration. After data revision, the Exclude checkbox can be disabled to consider the revised line during test group import.

• Continue this review until all required lines are correct, complete, and ready for import.

Import the reviewed data

When the document header status is AI response succeeded, the Import button becomes available in the Document Intelligence staging form. Click Import to create a new test group in . The system creates a new test group header and one test group line for each staging line that is not excluded (Exclude = No). The created test group is imported as unapproved and inactive, so users must still complete the standard business review, approval, and activation steps before the test group can be used operationally.

Cancel a processing instance

If the document should no longer be processed, users can click Cancel in the Document Intelligence staging form. After confirmation, the header status is updated to Canceled. Background processing in the AI application is not stopped, but  does not continue the import flow for that instance. If an AI response is received after cancellation, the instance remains shown as Canceled.

Review logs and history

Use the Log button in the Document Intelligence staging form to open the Document Intelligence log form and review the detailed sequence of processing events, status changes, and related messages for the selected document.

You can use the following checkboxes in the Document Intelligence staging form to view more lines:

• Show completed: if enabled, shows lines in Import completed or Canceled status.

• Show all users: if enabled, shows records for documents imported by other users.

To review previously processed documents in a read-only overview, open System administration > Inquiries > Document Intelligence > Document import history. This form displays all import instances, including completed and canceled records.

Important notes and limitations

• Only .pdf, .doc and .docx files are supported.

• The test group data should be formatted within a table format in the uploaded document.

• Only the following test types and test group fields are currently supported for import:

  • Fraction or Integer test type:

    • Min and Max fields under the Result range field group.

    • Standard, Min, and Max fields under the Result range field group.

  • Option test type:

    • Variable and Default outcome fields under the Test variables field group.

Any other field must be manually updated after import. Tests of type Criterion can still be proposed as matching tests, but the Criterion field will be left empty and available for user update.

• The current functionality supports the import of one test group per document. If a document contains multiple test groups, split the content into separate files and process them individually.

• If processing fails before the staging data is created, users must restart with a new upload. If import fails after staging data has been created, users can review and correct the staging information and then retry the import.

Troubleshooting

Use the guidance in the table when the document import process does not behave as expected. We recommend canceling the failing lines via the Cancel button to clean up the Document Intelligence staging form (such records will remain visible when enabling the Show completed checkbox).